zrwangems/dsh-zrwang-plugin / ars/skills/academic-pipeline
academic-pipeline
Flagged学术研究总编排:研究→写作→诚信检查→评审→返修→复审→终检→定稿的十阶段流水线,每阶段停在用户确认点,含材料护照与 7 类 AI 研究失败模式检查。Triggers: academic pipeline, research to paper, full paper workflow, end-to-end paper | 完整论文流程, 从研究到论文, 端到端写论文, 论文流水线。
_shared/contracts/passport/audit_artifact_entry.schema.json
View on GitHub{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"$id": "https://github.com/Imbad0202/academic-research-skills/shared/contracts/passport/audit_artifact_entry.schema.json",
"title": "Material Passport Audit Artifact Entry",
"description": "One entry in Material Passport Schema 9 audit_artifact[]. Records one cross-model audit run for one downstream agent's deliverable. Two lifecycle states share this schema via oneOf: 'proposal' (wrapper-emitted, pre-verification) and 'persisted' (orchestrator-merged, post-verification). See docs/design/2026-04-30-ars-v3.6.7-step-6-orchestrator-hooks-spec.md §3.1-§3.2 and §3.7 for cross-artifact invariants. Cross-field rules (PASS/MINOR/MATERIAL/AUDIT_FAILED counts; acknowledgement scoping; etc.) are lint-enforced in scripts/check_audit_artifact_consistency.py (Phase 6.3), NOT in this schema.",
"type": "object",
"additionalProperties": false,
"required": [
"stage",
"agent",
"deliverable_path",
"deliverable_sha",
"run_id",
"bundle_manifest_sha",
"artifact_paths",
"verdict"
],
"properties": {
"stage": {
"type": "integer",
"minimum": 1,
"maximum": 6,
"description": "Destination stage the just-completed deliverable is about to enter (per §5.1). v3.6.7 hooks emit stage 2 for synthesis_agent / research_architect_agent exits and stage 5 for report_compiler_agent exit."
},
"agent": {
"type": "string",
"enum": [
"synthesis_agent",
"research_architect_agent",
"report_compiler_agent"
],
"description": "One of the three v3.6.7 audited agents. Other agent names are reserved for future versions and rejected by Schema 9 lint."
},
"deliverable_path": {
"$ref": "#/$defs/repo_relative_path",
"description": "Repo-relative POSIX path to the audited deliverable. No leading '/' and no '..' segments."
},
"deliverable_sha": {
"$ref": "#/$defs/sha256_hex",
"description": "SHA-256 (lowercase hex) of the deliverable file at the moment audit was dispatched. Used by orchestrator to detect post-audit deliverable mutation."
},
"run_id": {
"$ref": "#/$defs/run_id",
"description": "Canonical run identifier emitted by the wrapper. Format <ISO-8601-Z>-<4-hex>. Per §3.7 family F (F1)."
},
"bundle_id": {
"type": "string",
"minLength": 1,
"description": "Optional opaque tag shared by all entries that belong to one logical multi-file audit bundle (§4.5). When present MUST equal the sidecar's prompt.bundle.bundle_id (§3.7 B9, lint-enforced)."
},
"bundle_manifest_sha": {
"$ref": "#/$defs/sha256_hex",
"description": "SHA-256 over the canonical bundle manifest of primary + supporting + template files (§3.6)."
},
"artifact_paths": {
"type": "object",
"additionalProperties": false,
"required": ["jsonl", "sidecar", "verdict"],
"properties": {
"jsonl": {
"allOf": [
{ "$ref": "#/$defs/repo_relative_path" },
{ "type": "string", "pattern": "\\.jsonl$" }
]
},
"sidecar": {
"allOf": [
{ "$ref": "#/$defs/repo_relative_path" },
{ "type": "string", "pattern": "\\.meta\\.json$" }
]
},
"verdict": {
"allOf": [
{ "$ref": "#/$defs/repo_relative_path" },
{ "type": "string", "pattern": "\\.verdict\\.yaml$" }
]
}
},
"description": "All three paths are repo-relative POSIX (no leading '/', no '..' segments). Path B uses these to locate JSONL/sidecar/verdict evidence; an absolute or escaping path would let a forged proposal point verification outside the repo."
},
"verdict": { "$ref": "#/$defs/verdict_block" },
"acknowledgement": { "$ref": "#/$defs/acknowledgement_block" }
},
"allOf": [
{
"title": "[A2] AUDIT_FAILED requires verdict.failure_reason; mirrors audit_verdict.schema.json",
"description": "Entry-side mirror of the verdict-file rule (§3.7 A2). Without this, a wrapper-emitted AUDIT_FAILED proposal missing failure_reason passes Path B4 schema validation; Path B5 then short-circuits and tries to surface a non-existent reason in the BLOCK message. Mirror keeps the rule enforceable at the first artifact orchestrator reads (entry), not just at the verdict file.",
"if": {
"properties": {
"verdict": {
"properties": { "status": { "const": "AUDIT_FAILED" } },
"required": ["status"]
}
},
"required": ["verdict"]
},
"then": {
"properties": {
"verdict": { "required": ["failure_reason"] }
}
}
},
{
"title": "[A2 inverse] non-AUDIT_FAILED proposals MUST NOT carry failure_reason",
"if": {
"properties": {
"verdict": {
"properties": {
"status": { "type": "string", "enum": ["PASS", "MINOR", "MATERIAL"] }
},
"required": ["status"]
}
},
"required": ["verdict"]
},
"then": {
"properties": {
"verdict": { "not": { "required": ["failure_reason"] } }
}
}
}
],
"oneOf": [
{
"title": "proposal",
"description": "Wrapper-emitted, pre-verification. verdict.verified_at and verdict.verified_by MUST be absent (Pattern C3 attack surface defense per §3.2 / §3.7 E3+E4). acknowledgement is forbidden — orchestrator-only write per §5.4 / §3.7 A4.",
"not": { "required": ["acknowledgement"] },
"properties": {
"verdict": {
"type": "object",
"not": {
"anyOf": [
{ "required": ["verified_at"] },
{ "required": ["verified_by"] }
]
}
}
}
},
{
"title": "persisted",
"description": "Orchestrator-merged, post-verification. AUDIT_FAILED excluded — see §3.2 'Why persisted excludes AUDIT_FAILED' rationale. acknowledgement is allowed only on MATERIAL entries — schema-level if/then below blocks the hand-edit attack where someone adds an acknowledgement block to a non-MATERIAL persisted entry to claim residue acknowledgement that was never solicited (§3.7 A4).",
"properties": {
"verdict": {
"type": "object",
"required": ["verified_at", "verified_by"],
"properties": {
"status": { "enum": ["PASS", "MINOR", "MATERIAL"] },
"verified_at": {
"$ref": "#/$defs/rfc3339_ms_utc",
"description": "RFC 3339 UTC datetime, millisecond precision (mirrors §3.4 sidecar timing precision so D1 latest-by-verified_at ordering has resolution >= 1 ms; see §5.4 monotonic-bump rule)."
},
"verified_by": {
"const": "pipeline_orchestrator_agent"
}
}
}
},
"allOf": [
{
"title": "[A4] acknowledgement requires verdict.status == MATERIAL",
"if": {
"properties": {
"verdict": {
"properties": { "status": { "enum": ["PASS", "MINOR"] } },
"required": ["status"]
}
},
"required": ["verdict"]
},
"then": {
"not": { "required": ["acknowledgement"] }
}
}
]
}
],
"$defs": {
"repo_relative_path": {
"type": "string",
"minLength": 1,
"pattern": "^[^/][^\\s]*$",
"not": { "pattern": "(^|/)\\.\\.(/|$)" },
"description": "Repo-relative POSIX path (no leading '/', no '..' segments). Sibling audit/* schemas duplicate this pattern; keep the regex pair in sync. Layer 3 lint reads these paths from disk for SHA-256 / git_sha verification — accepting absolute or escaping paths would let a forged proposal hash arbitrary files."
},
"sha256_hex": {
"type": "string",
"pattern": "^[a-f0-9]{64}$",
"description": "Lowercase hex SHA-256 digest. Sibling audit/* schemas duplicate this pattern (cross-file $ref is operationally awkward without a registry resolver); keep regex in sync via §3.7 family F naming-convention review."
},
"run_id": {
"type": "string",
"pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}-[0-9]{2}-[0-9]{2}Z-[0-9a-f]{4}$",
"description": "<ISO-8601-Z>-<4-hex>. F1 in §3.7 naming family. Sibling audit/* schemas duplicate this pattern."
},
"rfc3339_ms_utc": {
"type": "string",
"format": "date-time",
"pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}\\.[0-9]{3}Z$",
"description": "RFC 3339 UTC datetime with millisecond precision. Format keyword is advisory (validator must enable format_checker); regex is the load-bearing constraint."
},
"verdict_block": {
"type": "object",
"additionalProperties": false,
"required": ["status", "round", "target_rounds", "finding_counts"],
"properties": {
"status": {
"type": "string",
"enum": ["PASS", "MINOR", "MATERIAL", "AUDIT_FAILED"],
"description": "Proposal-arm enum; persisted arm narrows to {PASS,MINOR,MATERIAL} via oneOf override (§3.2 Lifecycle-conditional fields)."
},
"failure_reason": {
"type": "string",
"minLength": 1,
"maxLength": 500,
"pattern": "^[^\\n\\r]+$",
"description": "One-line human-readable reason. Required iff status == AUDIT_FAILED (cross-field rule, lint-enforced per §3.7 A2)."
},
"round": { "type": "integer", "minimum": 1 },
"target_rounds": { "type": "integer", "minimum": 1 },
"finding_counts": {
"type": "object",
"additionalProperties": false,
"required": ["p1", "p2", "p3"],
"properties": {
"p1": { "type": "integer", "minimum": 0 },
"p2": { "type": "integer", "minimum": 0 },
"p3": { "type": "integer", "minimum": 0 }
}
},
"verified_at": {
"type": "string",
"format": "date-time"
},
"verified_by": {
"type": "string",
"minLength": 1
}
}
},
"acknowledgement_block": {
"type": "object",
"additionalProperties": false,
"required": ["finding_ids", "acknowledged_at", "acknowledged_by"],
"properties": {
"finding_ids": {
"type": "array",
"minItems": 1,
"items": { "type": "string", "minLength": 1 },
"description": "Non-empty array; full coverage of every current findings[].id is enforced by lint (§3.7 B10)."
},
"acknowledged_at": {
"$ref": "#/$defs/rfc3339_ms_utc",
"description": "RFC 3339 UTC, millisecond precision. Equals verdict.verified_at by construction (§3.7 B8)."
},
"acknowledged_by": { "const": "user" }
}
}
}
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |