Skip to content

hotskills.dev

Not in the top 1,000 by GitHub stars

HKUST-KnowComp/DeepRefine-Skill

An agent skill to evolve the quality of LLM-Wiki (Graphify) at test time.

Gemini CLI Testing

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

`SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Local scripts

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`deeprefine_skill/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Network or broad shell

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`deeprefine_skill/claude_skill/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`deeprefine_skill/codex_skill/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`deeprefine_skill/dsh_skill/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`deeprefine_skill/gemini_extension/skills/deeprefine/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

`skills/deeprefine/SKILL.md`

Week of Oct 26, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → 9b17e7c086af

No earlier version was observed for comparison.

Compare on GitHub

View on GitHub Add to my skills

Verified commit 9b17e7c086af8fd30b023868b1e96ae3295b8f04

Version-pinned installation · Gemini CLI

git clone --filter=blob:none --no-checkout https://github.com/HKUST-KnowComp/DeepRefine-Skill.git .hotskills/hkust-knowcomp-deeprefine-skill && git -C .hotskills/hkust-knowcomp-deeprefine-skill checkout --detach 9b17e7c086af8fd30b023868b1e96ae3295b8f04
# Copy each selected skill directory from .hotskills/hkust-knowcomp-deeprefine-skill into .gemini/skills/<skill-name>/

90GitHub stars

-1star in 7 days

7skills in this repository

0commits in 30 days

Heat

0.2

Star growth, last 7 daysNot available
Star growth, last 24 hoursNot available
Rank movement, last 7 daysNot available
Commits, last 30 daysNot available
Accounts asking about it through MCP, last 7 daysNot available
Hot votes this weekNot available

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

30-day change not yet available: not enough history.

Individual skills in this repository

deeprefine /
deeprefine deeprefine_skill
deeprefine deeprefine_skill/claude_skill
deeprefine deeprefine_skill/codex_skill
deeprefine deeprefine_skill/dsh_skill
deeprefine deeprefine_skill/gemini_extension/skills/deeprefine
deeprefine skills/deeprefine

How it works

Choose a skill

deeprefine

When your agent uses it

Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.

What it loads into context

Frontmatter

57 tokens

Always available to the agent

SKILL.md

4,072 tokens

When the skill is activated

Reference documents

17,713 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Default safety policy: dry-run only
FORBIDDEN (hard stop)
Constants (match `refine_runner.py` / DeepRefine)
Mandatory artifact
Query queue selection (default behavior of `/deeprefine`)
Control flow (must match `DeepRefine.refine()`)
Evidence-aware review rules
LLM prompts (verbatim — do not paraphrase)
Per-query checklist (report in chat)
Commands (in order)
`loop_trace_*.json` schema
Optional: CLI mode (FAISS + API)
Paths

Scripts executed separately

eval/benchmarks/kg_structure.py
eval/benchmarks/refinement_recall.py
scripts/deeprefine.py
scripts/validate_graph_schema.py
tests/test_dsh_integration.py
tests/test_opencode_integration.py
tests/test_wiki_refresh.py

Files

.gitignoreasset603 bytes
GEMINI.mddoc1,641 bytes
LICENSEasset1,071 bytes
MANIFEST.inasset501 bytes
README.mddoc24,223 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
SKILL.mdcore16,592 bytes
assets/claudecode-color.pngasset1,467 bytes
assets/codex-color.pngasset7,578 bytes
assets/cursor_CUBE_25D.pngasset62,709 bytes
assets/demo-typing-deeprefine.svgasset6,643 bytes
assets/gemini-cli-icon_full-color@4x.pngasset866,495 bytes
assets/github-copilot__1_-removebg-preview.pngasset217,080 bytes
assets/harness.pngasset86,687 bytes
assets/icons.pngasset1,452,478 bytes
assets/icons2.pngasset269,804 bytes
assets/icons3.pngasset99,043 bytes
assets/opencode-logo-dark.pngasset2,491 bytes
assets/opencode-wordmark-dark.pngasset2,837 bytes
commands/deeprefine.tomlasset1,744 bytes
commands/deeprefine/apply.tomlasset1,289 bytes
commands/deeprefine/review.tomlasset1,373 bytes
docs/codex.mddoc2,537 bytes
docs/copilot-cli.mddoc3,705 bytes
docs/dsh.mddoc1,968 bytes
docs/feat-verification-fixtures.mddoc7,958 bytes
docs/gemini-cli.mddoc2,068 bytes
docs/phase-b-report.mddoc6,343 bytes
docs/verification-data.mddoc6,837 bytes
eval/README.mddoc7,494 bytes
eval/benchmarks/kg_structure.pyruns6,139 bytes
eval/benchmarks/refinement_recall.pyruns7,724 bytes
eval/smoke-tests/README.mddoc3,202 bytes
fixtures/refinement-ambiguous.txtdoc74 bytes
fixtures/refinement-good.txtdoc78 bytes
fixtures/refinement-invalid.txtdoc67 bytes
fixtures/synth-minimal.jsonasset5,042 bytes
fixtures/trace-invalid-order.jsonasset1,419 bytes
fixtures/trace-valid-early-exit.jsonasset1,098 bytes
fixtures/trace-valid.jsonasset2,908 bytes
gemini-extension.jsonasset252 bytes
pyproject.tomlasset2,114 bytes
scripts/deeprefine.pyruns429 bytes
scripts/validate_graph_schema.pyruns4,698 bytes
tests/test_dsh_integration.pyruns9,166 bytes
tests/test_opencode_integration.pyruns10,140 bytes
tests/test_wiki_refresh.pyruns13,424 bytes

Read SKILL.md


# DeepRefine — Agent refinement loop (strict)

## Default safety policy: dry-run only

A normal `/deeprefine` invocation **MUST NEVER** call `deeprefine apply`.

The default `/deeprefine` workflow must stop after:

1. `deeprefine loop validate`
2. `deeprefine review`
3. showing the proposed actions and HIGH/MEDIUM/LOW review report to the user

Then ask the user for explicit approval.

View the complete source on GitHub

deeprefine

When your agent uses it

Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.

What it loads into context

Frontmatter

57 tokens

Always available to the agent

SKILL.md

4,738 tokens

When the skill is activated

Reference documents

14,899 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Default safety policy: dry-run only
FORBIDDEN (hard stop)
Constants (match `refine_runner.py` / DeepRefine)
Mandatory artifact
Query queue selection (default behavior of `/deeprefine`)
Control flow (must match `DeepRefine.refine()`)
Evidence-aware review rules
LLM prompts (verbatim — do not paraphrase)
Per-query checklist (report in chat)
Commands (in order)
`loop_trace_*.json` schema
Optional: CLI mode (FAISS + API)
Paths
Checkpoint timeline (compare refinement chains)

Scripts executed separately

__init__.py
action_review.py
adapter_graphify.py
agent_graph.py
agent_loop.py
agent_prompts.py
cli.py
history.py
installers.py
obsidian_refresh.py
paths.py
refine_runner.py
wiki_importer.py
wiki_refresh.py
wiki_retrieval.py
wiki_update.py

Files

SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL_COPILOT.mddoc22,653 bytes
SKILL_OPENCODE.mddoc34,198 bytes
__init__.pyruns105 bytes
action_review.pyruns14,356 bytes
adapter_graphify.pyruns6,848 bytes
agent_graph.pyruns7,406 bytes
agent_loop.pyruns8,809 bytes
agent_prompts.pyruns2,995 bytes
cli.pyruns49,304 bytes
commands/opencode/deeprefine-apply.mddoc1,137 bytes
commands/opencode/deeprefine-review.mddoc892 bytes
commands/opencode/deeprefine.mddoc757 bytes
gemini_extension/GEMINI.mddoc1,641 bytes
gemini_extension/commands/deeprefine.tomlasset1,744 bytes
gemini_extension/commands/deeprefine/apply.tomlasset1,274 bytes
gemini_extension/commands/deeprefine/review.tomlasset1,373 bytes
gemini_extension/gemini-extension.jsonasset252 bytes
history.pyruns6,298 bytes
installers.pyruns23,897 bytes
obsidian_refresh.pyruns614 bytes
paths.pyruns6,884 bytes
refine_runner.pyruns9,424 bytes
wiki_importer.pyruns11,859 bytes
wiki_refresh.pyruns12,681 bytes
wiki_retrieval.pyruns6,513 bytes
wiki_update.pyruns15,102 bytes

Read SKILL.md


# DeepRefine — Agent refinement loop (strict)

## Default safety policy: dry-run only

A normal `/deeprefine` invocation **MUST NEVER** call `deeprefine apply`.

The default `/deeprefine` workflow must stop after:

1. `deeprefine loop validate`
2. `deeprefine review`
3. showing the proposed actions and HIGH/MEDIUM/LOW review report to the user

Then ask the user for explicit approval.

View the complete source on GitHub

deeprefine

When your agent uses it

Claude Code adapter for the DeepRefine agent-native refinement loop. Use when the user invokes /deeprefine, or asks to refine, diagnose, review, or apply changes to a Graphify / LLM-Wiki knowledge graph. Must follow the canonical DeepRefine skill rules and stop for review before graph writes.

What it loads into context

Frontmatter

105 tokens

Always available to the agent

SKILL.md

1,318 tokens

When the skill is activated

Reference documents

3,280 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Claude Code Invocation
Hard Safety Policy
Mode Selection
Non-Negotiable Rules
What to Load From References

Files

SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
SKILL.mdcore5,650 bytes
references/deeprefine-workflow.mddoc5,569 bytes
references/llm-prompts.mddoc3,339 bytes
references/trace-and-commands.mddoc4,333 bytes

Read SKILL.md


# DeepRefine - Claude Code Adapter

This file is the Claude Code-specific entrypoint. It keeps the platform rules
small and loads longer DeepRefine procedure details only when needed:

- Full workflow, queue selection, refinement branch logic, and review rules:
  [references/deeprefine-workflow.md](references/deeprefine-workflow.md)
- Verbatim judgement, abduction, and refinement prompts:
  [references/llm-prompts.md](references/llm-prompts.md)
- Checklist, command sequence, trace schema, paths, and CLI mode:
  [references/trace-and-commands.md](references/trace-and-commands.md)

Do not reimplement or shorten the algorithm from memory. Load the relevant
reference file before executing that part of the workflow.

View the complete source on GitHub

deeprefine

When your agent uses it

Codex adapter for the DeepRefine agent-native refinement loop. Use when the user invokes $deeprefine or /deeprefine, or asks to refine, diagnose, review, or apply changes to a Graphify / LLM-Wiki knowledge graph. Must follow the canonical DeepRefine skill rules and stop for review before graph writes.

What it loads into context

Frontmatter

85 tokens

Always available to the agent

SKILL.md

1,323 tokens

When the skill is activated

Reference documents

3,504 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Codex Invocation
Hard Safety Policy
Mode Selection
Non-Negotiable Rules
What to Load From References

Files

SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
SKILL.mdcore5,616 bytes
agents/openai.yamlasset275 bytes
references/deeprefine-workflow.mddoc5,602 bytes
references/llm-prompts.mddoc3,339 bytes
references/trace-and-commands.mddoc5,250 bytes

Read SKILL.md


# DeepRefine - Codex Adapter

This file is the Codex-specific entrypoint. It keeps the platform rules small
and loads longer DeepRefine procedure details only when needed:

- Full workflow, queue selection, refinement branch logic, and review rules:
  [references/deeprefine-workflow.md](references/deeprefine-workflow.md)
- Verbatim judgement, abduction, and refinement prompts:
  [references/llm-prompts.md](references/llm-prompts.md)
- Checklist, command sequence, trace schema, paths, and CLI mode:
  [references/trace-and-commands.md](references/trace-and-commands.md)

Do not reimplement or shorten the algorithm from memory. Load the relevant
reference file before executing that part of the workflow.

View the complete source on GitHub

deeprefine

When your agent uses it

DeepSeek Harness (dsh) adapter for the DeepRefine agent-native refinement loop. Use when the user invokes /deeprefine, or asks to refine, diagnose, review, or apply changes to a Graphify / LLM-Wiki knowledge graph. Must follow the canonical DeepRefine skill rules and stop for review before graph writes.

What it loads into context

Frontmatter

94 tokens

Always available to the agent

SKILL.md

1,526 tokens

When the skill is activated

Reference documents

3,302 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

dsh Invocation
Hard Safety Policy
Mode Selection
Non-Negotiable Rules
What to Load From References

Files

SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
SKILL.mdcore6,483 bytes
references/deeprefine-workflow.mddoc5,661 bytes
references/llm-prompts.mddoc3,344 bytes
references/trace-and-commands.mddoc4,333 bytes

Read SKILL.md


# DeepRefine - DeepSeek Harness (dsh) Adapter

This file is the dsh-specific entrypoint. It keeps the platform rules small
and loads longer DeepRefine procedure details only when needed:

- Full workflow, queue selection, refinement branch logic, and review rules:
  [references/deeprefine-workflow.md](references/deeprefine-workflow.md)
- Verbatim judgement, abduction, and refinement prompts:
  [references/llm-prompts.md](references/llm-prompts.md)
- Checklist, command sequence, trace schema, paths, and CLI mode:
  [references/trace-and-commands.md](references/trace-and-commands.md)

Do not reimplement or shorten the algorithm from memory. Load the relevant
reference file before executing that part of the workflow.

View the complete source on GitHub

deeprefine

When your agent uses it

Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.

What it loads into context

Frontmatter

57 tokens

Always available to the agent

SKILL.md

4,738 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Default safety policy: dry-run only
FORBIDDEN (hard stop)
Constants (match `refine_runner.py` / DeepRefine)
Mandatory artifact
Query queue selection (default behavior of `/deeprefine`)
Control flow (must match `DeepRefine.refine()`)
Evidence-aware review rules
LLM prompts (verbatim — do not paraphrase)
Per-query checklist (report in chat)
Commands (in order)
`loop_trace_*.json` schema
Optional: CLI mode (FAISS + API)
Paths
Checkpoint timeline (compare refinement chains)

Files

SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes

Read SKILL.md


# DeepRefine — Agent refinement loop (strict)

## Default safety policy: dry-run only

A normal `/deeprefine` invocation **MUST NEVER** call `deeprefine apply`.

The default `/deeprefine` workflow must stop after:

1. `deeprefine loop validate`
2. `deeprefine review`
3. showing the proposed actions and HIGH/MEDIUM/LOW review report to the user

Then ask the user for explicit approval.

View the complete source on GitHub

deeprefine

When your agent uses it

Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.

What it loads into context

Frontmatter

57 tokens

Always available to the agent

SKILL.md

4,738 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

Default safety policy: dry-run only
FORBIDDEN (hard stop)
Constants (match `refine_runner.py` / DeepRefine)
Mandatory artifact
Query queue selection (default behavior of `/deeprefine`)
Control flow (must match `DeepRefine.refine()`)
Evidence-aware review rules
LLM prompts (verbatim — do not paraphrase)
Per-query checklist (report in chat)
Commands (in order)
`loop_trace_*.json` schema
Optional: CLI mode (FAISS + API)
Paths
Checkpoint timeline (compare refinement chains)

Files

SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes
SKILL.mdcore19,251 bytes

Read SKILL.md


# DeepRefine — Agent refinement loop (strict)

## Default safety policy: dry-run only

A normal `/deeprefine` invocation **MUST NEVER** call `deeprefine apply`.

The default `/deeprefine` workflow must stop after:

1. `deeprefine loop validate`
2. `deeprefine review`
3. showing the proposed actions and HIGH/MEDIUM/LOW review report to the user

Then ask the user for explicit approval.

View the complete source on GitHub

Skills included

deeprefine
deeprefine
deeprefine
deeprefine
deeprefine
deeprefine
deeprefine

Repository facts

License: MIT
Created: Jun 1, 2026
Last pushed: Aug 23, 2026
Forks: 9
Open issues: 1
Project activity: Slowing

Before you install

Network or broad shell

Automated signals, not a guarantee

Checked Nov 6, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 55 notes

SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/claude_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/claude_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/claude_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
deeprefine_skill/claude_skill/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed

1 broad permissionin 1 file: deeprefine_skill/SKILL_COPILOT.md:1
2 dangerous shell commandsin 2 files: docs/feat-verification-fixtures.md:135 docs/verification-data.md:143
18 network referencesin 3 files: README.md:21 pyproject.toml:31 assets/demo-typing-deeprefine.svg:1
No hidden or obfuscated text
No instruction overrides
23 executable scriptsin 23 files: deeprefine_skill/__init__.py:1 deeprefine_skill/action_review.py:1 deeprefine_skill/adapter_graphify.py:1 and 20 more
28 references to secretsin 7 files: deeprefine_skill/paths.py:23 deeprefine_skill/refine_runner.py:65 eval/benchmarks/refinement_recall.py:102 and 4 more
No requests to vote in Hot

Show all 72 signals by file

`README.md`

Network accessline 21

[![PyPi](https://img.shields.io/badge/PyPi-v0.2.0-blue.svg)](https://pypi.org/project/deeprefine-cli/0.2.0/)

Network accessline 22

[![Python](https://img.shields.io/badge/Python-3.10,3.11,3.12-blue.svg)](https://pypi.org/project/deeprefine-cli/0.2.0/)

Network accessline 23

[![Paper](https://img.shields.io/badge/Paper-DeepRefine-b31b1b.svg)](https://arxiv.org/pdf/2605.10488)

Network accessline 24

[![Project](https://img.shields.io/badge/Project-DeepRefine-green.svg)](https://github.com/HKUST-KnowComp/DeepRefine)

Network accessline 28

DeepRefine-Skill plugs into agent workflows and use a single command `/deeprefine` in your agent (Cursor, Copilot CLI, Gemini CLI, Codex, OpenCode, Claude Code) to refine and evolve your LLM-Wiki (e.g., **[graphify](https://github.com/safis

Network accessline 38

<p>   <a href="https://cursor.com" title="Cursor"><img src="./assets/cursor_CUBE_25D.png" alt="Cursor" height="40"/></a>&nbsp;&nbsp;

Network accessline 39

<a href="https://cursor.com" title="Cursor"><img src="./assets/cursor_CUBE_25D.png" alt="Cursor" height="40"/></a>&nbsp;&nbsp;

Network accessline 40

<a href="https://github.com/google-gemini/gemini-cli" title="Gemini CLI"><img src="./assets/gemini-cli-icon_full-color@4x.png" alt="Gemini CLI" height="40"/></a>&nbsp;&nbsp;

Network accessline 41

<a href="https://docs.github.com/en/copilot" title="GitHub Copilot CLI" style="text-decoration: none; color: inherit;"><img src="./assets/github-copilot__1_-removebg-preview.png" height="40" style="vertical-align: middle;" alt="Copilot Icon

Network accessline 42

<a href="https://openai.com/codex/" title="Codex CLI" style="text-decoration: none; color: inherit;"><img src="./assets/codex-color.png" height="40" style="vertical-align: middle;" alt="Copilot Icon" /></a>&nbsp;&nbsp;

Network accessline 43

<a href="https://opencode.ai/" title="OpenCode" style="text-decoration: none; color: inherit;"><img src="./assets/opencode-logo-dark.png" height="40" style="vertical-align: middle;" alt="Copilot Icon" /></a>&nbsp;&nbsp;

Network accessline 44

<a href="https://claude.com/product/claude-code" title="Claude Code" style="text-decoration: none; color: inherit;"><img src="./assets/claudecode-color.png" height="40" style="vertical-align: middle;" alt="Copilot Icon" /></a>&nbsp;&nbsp;

Access to secretsline 521
```
[Credential-related source omitted]
```
Access to secretsline 522
```
[Credential-related source omitted]
```
Access to secretsline 523
```
[Credential-related source omitted]
```

`assets/demo-typing-deeprefine.svg`

Network accessline 1

<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 560 60" width="560" height="60">

`deeprefine_skill/SKILL_COPILOT.md`

Broad permissionsline 1
```
shell
```

`deeprefine_skill/init.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/action_review.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/adapter_graphify.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/agent_graph.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/agent_loop.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/agent_prompts.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/cli.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/history.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/installers.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/obsidian_refresh.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/paths.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 23
```
[Credential-related source omitted]
```
Access to secretsline 176
```
[Credential-related source omitted]
```
Access to secretsline 177
```
[Credential-related source omitted]
```
Access to secretsline 181
```
[Credential-related source omitted]
```
Access to secretsline 182
```
[Credential-related source omitted]
```
Access to secretsline 183
```
[Credential-related source omitted]
```
Access to secretsline 184
```
[Credential-related source omitted]
```
Access to secretsline 186
```
[Credential-related source omitted]
```
Access to secretsline 187
```
[Credential-related source omitted]
```
Access to secretsline 189
```
[Credential-related source omitted]
```
Access to secretsline 192
```
[Credential-related source omitted]
```

`deeprefine_skill/refine_runner.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 65
```
[Credential-related source omitted]
```
Access to secretsline 69
```
[Credential-related source omitted]
```
Access to secretsline 70
```
[Credential-related source omitted]
```
Access to secretsline 71
```
[Credential-related source omitted]
```
Access to secretsline 73
```
[Credential-related source omitted]
```
Access to secretsline 80
```
[Credential-related source omitted]
```
Access to secretsline 84
```
[Credential-related source omitted]
```

`deeprefine_skill/wiki_importer.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/wiki_refresh.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 238
```
[Credential-related source omitted]
```

`deeprefine_skill/wiki_retrieval.py`

Executable scriptline 1
```
Executable file
```

`deeprefine_skill/wiki_update.py`

Executable scriptline 1
```
Executable file
```

`docs/feat-verification-fixtures.md`

Dangerous shell operationline 135
```
rm -rf /tmp/test_kb
```

`docs/verification-data.md`

Dangerous shell operationline 143
```
rm -rf /tmp/test_kb
```

`eval/README.md`

Access to secretsline 91
```
[Credential-related source omitted]
```

`eval/benchmarks/kg_structure.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```

`eval/benchmarks/refinement_recall.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```
Access to secretsline 102
```
[Credential-related source omitted]
```
Access to secretsline 203
```
[Credential-related source omitted]
```
Access to secretsline 204
```
[Credential-related source omitted]
```
Access to secretsline 212
```
[Credential-related source omitted]
```

`pyproject.toml`

Network accessline 31

[project.urls] Homepage = "https://github.com/HKUST-KnowComp/DeepRefine-Skill"

Network accessline 32

Homepage = "https://github.com/HKUST-KnowComp/DeepRefine-Skill"

Network accessline 33

Documentation = "https://github.com/HKUST-KnowComp/DeepRefine-Skill#readme"

Network accessline 34

Repository = "https://github.com/HKUST-KnowComp/DeepRefine-Skill"

Network accessline 35

Issues = "https://github.com/HKUST-KnowComp/DeepRefine-Skill/issues"

`scripts/deeprefine.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```

`scripts/validate_graph_schema.py`

Executable scriptline 1
```
#!/usr/bin/env python3
```

`tests/test_dsh_integration.py`

Executable scriptline 1
```
Executable file
```

`tests/test_opencode_integration.py`

Executable scriptline 1
```
Executable file
```
Access to secretsline 50
```
[Credential-related source omitted]
```

`tests/test_wiki_refresh.py`

Executable scriptline 1
```
Executable file
```

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Some files could not be read; the facts below still hold.

Signal confidence: 66 facts

Network access in prose of documentationREADME.md:21Low confidence
Network access in prose of documentationREADME.md:22Low confidence
Network access in prose of documentationREADME.md:23Low confidence
Network access in prose of documentationREADME.md:24Low confidence
Network access in prose of documentationREADME.md:28Low confidence
Network access in prose of documentationREADME.md:38Low confidence
Network access in prose of documentationREADME.md:39Low confidence
Network access in prose of documentationREADME.md:40Low confidence
Network access in prose of documentationREADME.md:41Low confidence
Network access in prose of documentationREADME.md:42Low confidence
Network access in prose of documentationREADME.md:43Low confidence
Network access in prose of documentationREADME.md:44Low confidence
Access to secrets in prose of documentationREADME.md:521Low confidence
Access to secrets in prose of documentationREADME.md:522Low confidence
Access to secrets in prose of documentationREADME.md:523Low confidence
Broad permissions in prose of documentationdeeprefine_skill/SKILL_COPILOT.md:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/__init__.py:1Low confidence
Executable script in code of a file that runsdeeprefine_skill/action_review.py:1High confidence
Executable script in code of a file that runsdeeprefine_skill/adapter_graphify.py:1High confidence
Executable script in prose of a file that runsdeeprefine_skill/agent_graph.py:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/agent_loop.py:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/agent_prompts.py:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/cli.py:1Low confidence
Executable script in code of a file that runsdeeprefine_skill/history.py:1High confidence
Executable script in prose of a file that runsdeeprefine_skill/installers.py:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/obsidian_refresh.py:1Low confidence
Executable script in code of a file that runsdeeprefine_skill/paths.py:1High confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:23Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:176Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:177Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:181Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:182Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:183Low confidence
Access to secrets in code of a file that runsdeeprefine_skill/paths.py:184High confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:186Low confidence
Access to secrets in code of a file that runsdeeprefine_skill/paths.py:187High confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:189Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/paths.py:192Low confidence
Executable script in code of a file that runsdeeprefine_skill/refine_runner.py:1High confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/refine_runner.py:65Low confidence
Access to secrets in a comment of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/refine_runner.py:69Low confidence
Access to secrets in code of a file that runsdeeprefine_skill/refine_runner.py:70High confidence
Access to secrets in code of a file that runsdeeprefine_skill/refine_runner.py:71High confidence
Access to secrets in code of a file that runsdeeprefine_skill/refine_runner.py:73High confidence
Access to secrets in code of a file that runsdeeprefine_skill/refine_runner.py:80High confidence
Access to secrets in code of a file that runsdeeprefine_skill/refine_runner.py:84High confidence
Executable script in prose of a file that runsdeeprefine_skill/wiki_importer.py:1Low confidence
Executable script in prose of a file that runsdeeprefine_skill/wiki_refresh.py:1Low confidence
Access to secrets in code of a file that runs: a variable, placeholder or call, not a stored valuedeeprefine_skill/wiki_refresh.py:238Low confidence
Executable script in prose of a file that runsdeeprefine_skill/wiki_retrieval.py:1Low confidence

and 16 more

Works with

Agent compatibility, from the files in this repository
Agent	Status	Evidence	Install to
Claude Code	Native	Frontmatter field disable-model-invocation in SKILL.md	`.claude/skills/<skill-name>/`
Codex	Compatible	Standard SKILL.md SKILL.md	`.agents/skills/<skill-name>/`
Cursor	Compatible	Standard SKILL.md SKILL.md	`.cursor/skills/<skill-name>/`
Gemini CLI	Native	Manifest deeprefine_skill/gemini_extension/gemini-extension.json	`.gemini/skills/<skill-name>/`
GitHub Copilot	Compatible	Standard SKILL.md SKILL.md	`.github/skills/<skill-name>/`
OpenCode	Compatible	Standard SKILL.md SKILL.md	`.opencode/skills/<skill-name>/`

Native The repository ships a manifest, folder or frontmatter field for this agent.
Compatible A standard SKILL.md that this agent can read.
Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.