Extract per-subsection “anchor facts” (NO PROSE) from evidence packs so the writer is forced to include concrete numbers/benchmarks/limitations instead of generic summaries.
=3 limitation/failure hooks (concrete, not generic “future work”)
If digits exist in the evidence pack: include >=1 cited numeric anchor (digit + citation in the same paragraph)
Note:
Anchor text is trimmed for readability and does not include ellipsis markers (to reduce accidental leakage into prose).
Script
Quick Start
uv run python .codex/skills/anchor-sheet/scripts/run.py --help
uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace>
All Options
--workspace <dir>
--unit-id <U###>
--inputs <semicolon-separated>
--outputs <semicolon-separated>
--checkpoint <C#>
Examples
Default IO:
uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace>
Explicit IO:
uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace> --inputs "outline/evidence_drafts.jsonl;citations/ref.bib" --outputs "outline/anchor_sheet.jsonl"
Refinement marker (recommended; prevents churn)
When you are satisfied with anchor facts (and they are actually subsection-specific), create:
outline/anchor_sheet.refined.ok
This is an explicit "I reviewed/refined this" signal:
prevents scripts from regenerating and undoing your work
(in strict runs) can be used as a completion signal before writing
1---2name: anchor-sheet3description: Extract per-subsection “anchor facts” (NO PROSE) from evidence packs so the writer is forced to include concrete numbers/benchmarks/limitations instead of generic summaries.4---56# Anchor Sheet (evidence → write hooks) [NO PROSE]78## Triggers & routing910- **Trigger**: anchor sheet, anchor facts, numeric anchors, evidence hooks, 写作锚点, 数字锚点, 证据钩子.11- **Use when**: `outline/evidence_drafts.jsonl` exists and you want stronger, evidence-anchored writing in `sections/*.md`.121314Purpose: make “what to actually say” explicit:15- select quantitative snippets (numbers/percentages)16- select evaluation anchors (benchmarks/datasets/metrics)17- select limitations/failure hooks1819This prevents the writer from producing paragraph-shaped but **content-poor** prose.2021## Inputs2223- `outline/evidence_drafts.jsonl`24- `citations/ref.bib`2526## Outputs2728- `outline/anchor_sheet.jsonl`2930## Output format (`outline/anchor_sheet.jsonl`)3132JSONL (one object per H3 subsection).3334Required fields:35- `sub_id`, `title`36- `anchors` (list; each anchor has `hook_type`, `text`, `citations`, and optional `paper_id/evidence_id/pointer`)3738## Workflow39401. Read `outline/evidence_drafts.jsonl`.412. Prefer anchors that contain:42 - a number (%, counts, scores)43 - an explicit benchmark/dataset/metric name44 - an explicit limitation/failure statement453. Filter anchors to only citation keys present in `citations/ref.bib`.464. Write `outline/anchor_sheet.jsonl`.4748## Quality checklist4950- [ ] Every H3 has >=10 cite-backed anchors (A150++ hard target).51- [ ] At least 1 anchor contains digits when the evidence pack contains digits.52- [ ] No placeholders (`TODO`/`…`/`(placeholder)`).5354## Consumption policy (for C5 writers)5556Anchors are intended to prevent “long but empty” prose. Treat them as **must-use hooks**, not optional ideas.5758Recommended minimums per H3 (A150++):59- >=3 protocol anchors (benchmark/dataset/metric/budget/tool access)60- >=3 limitation/failure hooks (concrete, not generic “future work”)61- If digits exist in the evidence pack: include >=1 cited numeric anchor (digit + citation in the same paragraph)6263Note:64- Anchor text is trimmed for readability and **does not** include ellipsis markers (to reduce accidental leakage into prose).6566## Script6768### Quick Start6970- `uv run python .codex/skills/anchor-sheet/scripts/run.py --help`71- `uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace>`7273### All Options7475- `--workspace <dir>`76- `--unit-id <U###>`77- `--inputs <semicolon-separated>`78- `--outputs <semicolon-separated>`79- `--checkpoint <C#>`8081### Examples8283- Default IO:84 - `uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace>`85- Explicit IO:86 - `uv run python .codex/skills/anchor-sheet/scripts/run.py --workspace <workspace> --inputs "outline/evidence_drafts.jsonl;citations/ref.bib" --outputs "outline/anchor_sheet.jsonl"`8788### Refinement marker (recommended; prevents churn)8990When you are satisfied with anchor facts (and they are actually subsection-specific), create:91- `outline/anchor_sheet.refined.ok`9293This is an explicit "I reviewed/refined this" signal:94- prevents scripts from regenerating and undoing your work95- (in strict runs) can be used as a completion signal before writing
Run npx skillmds add willoscar/anchor-sheet in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Extract per-subsection “anchor facts” (NO PROSE) from evidence packs so the writer is forced to include concrete numbers/benchmarks/limitations instead of generic summaries. It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. Capability flags: executes scripts. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
WILLOSCAR (@willoscar) published this skill. Their other Agent Skills are listed on their SkillMD profile.
WILLOSCAR