evidence-selfloop

Name: evidence-selfloop
Author: WILLOSCAR

489stars39forks

How to install evidence-selfloop?

willoscar/research-units-pipeline-skills/evidence-selfloop

$npx -y skills add WILLOSCAR/research-units-pipeline-skills --skill evidence-selfloop

Installs into the current project.

Summary

Evidence self-loop for surveys: read evidence bindings + evidence packs, then write an actionable upstream TODO plan (which stage/skill to fix) before writing more prose. Trigger: evidence self-loop, evidence loop, evidence gaps, binding gaps, blocking_missing, 证据自循环, 证据缺口回路.

Files · 1

View on GitHub

SKILL.md

1	# Evidence Self-loop (C3/C4 fix → rebind → redraft)
2
3	Purpose: make the evidence-first pipeline converge without writing filler prose.
4
5	This skill reads the intermediate evidence artifacts (briefs/bindings/packs) and produces an actionable TODO list that answers:
6
7	- Which subsections are under-supported?
8	- Is the problem mapping/coverage (C2) or evidence extraction (C3) or binding/planning (C4)?
9	- Which skill(s) should be rerun, in what order, to unblock high-quality writing?
10
11	## Inputs
12
13	- `outline/subsection_briefs.jsonl`
14	- `outline/evidence_bindings.jsonl` (expects `binding_gaps` / `binding_rationale` if available)
15	- `outline/evidence_drafts.jsonl` (expects `blocking_missing`, comparisons, eval protocol, limitations)
16	- Optional (improves routing):
17	- `outline/evidence_binding_report.md`
18	- `outline/anchor_sheet.jsonl`
19	- `papers/paper_notes.jsonl`
20	- `papers/fulltext_index.jsonl`
21	- `queries.md`
22
23	## Outputs
24
25	- `output/EVIDENCE_SELFLOOP_TODO.md` (report-class; always written)
26
27	## Self-loop contract (what “fixing evidence” means)
28
29	- Prefer fixing upstream evidence, not writing around gaps.
30	- If an evidence pack has `blocking_missing`, treat it as a STOP signal: strengthen notes/fulltext/mapping, then regenerate packs.
31	- If bindings show `binding_gaps`, treat it as a ROUTING signal: either enrich the evidence bank for the mapped papers, expand mapping coverage, or adjust `required_evidence_fields` if unrealistic.
32
33	Recommended rerun chain (minimal):
34
35	- If C3 evidence is thin: `pdf-text-extractor` → `paper-notes` → `evidence-binder` → `evidence-draft` → `anchor-sheet` → `writer-context-pack`
36	- If C2 coverage is weak: `section-mapper` → `outline-refiner` → (then rerun C3/C4 evidence skills)
37
38	## Workflow (analysis-only)
39
40	1) Read `queries.md` (if present)
41	- Use it only as a soft config hint (evidence_mode / draft_profile); do not override the artifact contract.
42
43	2) Read `outline/subsection_briefs.jsonl`
44	- For each `sub_id`, capture `axes` + `required_evidence_fields` (what evidence types this subsection expects).
45
46	3) Read `outline/evidence_bindings.jsonl`
47	- For each `sub_id`, surface `binding_rationale` and `binding_gaps` (what the binder could/could not cover from the evidence bank).
48
49	4) (Optional) Read `outline/evidence_binding_report.md`
50	- Use it as a human-readable summary; treat it as a view of `outline/evidence_bindings.jsonl`, not a separate truth source.
51
52	5) Read `outline/evidence_drafts.jsonl`
53	- Surface `blocking_missing` (STOP signals), and check for missing comparisons / eval protocol / limitations that would force hollow writing.
54
55	6) (Optional) Read `outline/anchor_sheet.jsonl`
56	- Check whether each subsection has at least a few citation-backed anchors (numbers / evaluation / limitations).
57
58	7) (Optional) Read `papers/paper_notes.jsonl` and `papers/fulltext_index.jsonl`
59	- Use these to route fixes: if evidence is abstract-only and missing eval tokens, prefer enriching notes/fulltext before drafting prose.
60
61	## What the report contains
62
63	- Summary counts: subsections with `blocking_missing`, with `binding_gaps`, and common failure reasons.
64	- Per-subsection TODO: the smallest upstream fix path (skills + artifacts) to make the subsection writeable.
65
66	## Status semantics (unblock rules)
67
68	This skill is the prewrite router for evidence quality. Treat its `Status:` line as the unblock contract:
69
70	- `PASS`: no `blocking_missing` and no `binding_gaps` -> proceed to C5 writing (but still scan non-blocking writability smells: low comparisons/eval/anchors often predict hollow prose).
71	- `OK`: no `blocking_missing`, but some `binding_gaps` -> you may draft, but expect weaker specificity; prefer fixing gaps first.
72	- `FAIL`: missing inputs OR any `blocking_missing` -> do not write filler prose; fix upstream and rerun C3/C4.
73
74	## Routing matrix (symptom -> root cause -> upstream fix)
75
76	Use this as a semantic routing table (not a script checklist). The goal is to fix the earliest broken intermediate artifact.
77
78	\| Symptom (where you see it) \| Likely root cause \| Inspect first \| Smallest upstream fix chain \|
79	\|---\|---\|---\|---\|
80	\| `evidence_drafts.blocking_missing: no usable citation keys` \| mapped papers lack `bibkey` / bibkeys not in `citations/ref.bib` \| `papers/paper_notes.jsonl` (bibkey fields), `citations/ref.bib` \| C3 `paper-notes` (ensure bibkeys) -> C4 `citation-verifier` -> rerun `evidence-binder` -> rerun `evidence-draft` \|
81	\| `blocking_missing: title-only evidence` \| retrieval/metadata lacks abstracts (or aggressive filtering) \| `papers/papers_raw.jsonl` abstracts, `papers/paper_notes.jsonl` evidence_level \| C1 `literature-engineer` (enrich metadata) OR C3 `pdf-text-extractor` (

People who liked this also viewed

6 picks

Type

evidence-selfloop

489stars39forks

View on GitHub

How to install evidence-selfloop?

willoscar/research-units-pipeline-skills/evidence-selfloop

$npx -y skills add WILLOSCAR/research-units-pipeline-skills --skill evidence-selfloop

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Evidence Self-loop (C3/C4 fix → rebind → redraft)
2
3	Purpose: make the evidence-first pipeline converge without writing filler prose.
4
5	This skill reads the intermediate evidence artifacts (briefs/bindings/packs) and produces an actionable TODO list that answers:
6
7	- Which subsections are under-supported?
8	- Is the problem mapping/coverage (C2) or evidence extraction (C3) or binding/planning (C4)?
9	- Which skill(s) should be rerun, in what order, to unblock high-quality writing?
10
11	## Inputs
12
13	- `outline/subsection_briefs.jsonl`
14	- `outline/evidence_bindings.jsonl` (expects `binding_gaps` / `binding_rationale` if available)
15	- `outline/evidence_drafts.jsonl` (expects `blocking_missing`, comparisons, eval protocol, limitations)
16	- Optional (improves routing):
17	- `outline/evidence_binding_report.md`
18	- `outline/anchor_sheet.jsonl`
19	- `papers/paper_notes.jsonl`
20	- `papers/fulltext_index.jsonl`
21	- `queries.md`
22
23	## Outputs
24
25	- `output/EVIDENCE_SELFLOOP_TODO.md` (report-class; always written)
26
27	## Self-loop contract (what “fixing evidence” means)
28
29	- Prefer fixing upstream evidence, not writing around gaps.
30	- If an evidence pack has `blocking_missing`, treat it as a STOP signal: strengthen notes/fulltext/mapping, then regenerate packs.
31	- If bindings show `binding_gaps`, treat it as a ROUTING signal: either enrich the evidence bank for the mapped papers, expand mapping coverage, or adjust `required_evidence_fields` if unrealistic.
32
33	Recommended rerun chain (minimal):
34
35	- If C3 evidence is thin: `pdf-text-extractor` → `paper-notes` → `evidence-binder` → `evidence-draft` → `anchor-sheet` → `writer-context-pack`
36	- If C2 coverage is weak: `section-mapper` → `outline-refiner` → (then rerun C3/C4 evidence skills)
37
38	## Workflow (analysis-only)
39
40	1) Read `queries.md` (if present)
41	- Use it only as a soft config hint (evidence_mode / draft_profile); do not override the artifact contract.
42
43	2) Read `outline/subsection_briefs.jsonl`
44	- For each `sub_id`, capture `axes` + `required_evidence_fields` (what evidence types this subsection expects).
45
46	3) Read `outline/evidence_bindings.jsonl`
47	- For each `sub_id`, surface `binding_rationale` and `binding_gaps` (what the binder could/could not cover from the evidence bank).
48
49	4) (Optional) Read `outline/evidence_binding_report.md`
50	- Use it as a human-readable summary; treat it as a view of `outline/evidence_bindings.jsonl`, not a separate truth source.
51
52	5) Read `outline/evidence_drafts.jsonl`
53	- Surface `blocking_missing` (STOP signals), and check for missing comparisons / eval protocol / limitations that would force hollow writing.
54
55	6) (Optional) Read `outline/anchor_sheet.jsonl`
56	- Check whether each subsection has at least a few citation-backed anchors (numbers / evaluation / limitations).
57
58	7) (Optional) Read `papers/paper_notes.jsonl` and `papers/fulltext_index.jsonl`
59	- Use these to route fixes: if evidence is abstract-only and missing eval tokens, prefer enriching notes/fulltext before drafting prose.
60
61	## What the report contains
62
63	- Summary counts: subsections with `blocking_missing`, with `binding_gaps`, and common failure reasons.
64	- Per-subsection TODO: the smallest upstream fix path (skills + artifacts) to make the subsection writeable.
65
66	## Status semantics (unblock rules)
67
68	This skill is the prewrite router for evidence quality. Treat its `Status:` line as the unblock contract:
69
70	- `PASS`: no `blocking_missing` and no `binding_gaps` -> proceed to C5 writing (but still scan non-blocking writability smells: low comparisons/eval/anchors often predict hollow prose).
71	- `OK`: no `blocking_missing`, but some `binding_gaps` -> you may draft, but expect weaker specificity; prefer fixing gaps first.
72	- `FAIL`: missing inputs OR any `blocking_missing` -> do not write filler prose; fix upstream and rerun C3/C4.
73
74	## Routing matrix (symptom -> root cause -> upstream fix)
75
76	Use this as a semantic routing table (not a script checklist). The goal is to fix the earliest broken intermediate artifact.
77
78	\| Symptom (where you see it) \| Likely root cause \| Inspect first \| Smallest upstream fix chain \|
79	\|---\|---\|---\|---\|
80	\| `evidence_drafts.blocking_missing: no usable citation keys` \| mapped papers lack `bibkey` / bibkeys not in `citations/ref.bib` \| `papers/paper_notes.jsonl` (bibkey fields), `citations/ref.bib` \| C3 `paper-notes` (ensure bibkeys) -> C4 `citation-verifier` -> rerun `evidence-binder` -> rerun `evidence-draft` \|
81	\| `blocking_missing: title-only evidence` \| retrieval/metadata lacks abstracts (or aggressive filtering) \| `papers/papers_raw.jsonl` abstracts, `papers/paper_notes.jsonl` evidence_level \| C1 `literature-engineer` (enrich metadata) OR C3 `pdf-text-extractor` (

People who liked this also viewed

6 picks

Type