# Fp Evidence Ledger

> Use only when routed by FP after execution, failure, or handoff to record canonical claims, checks, scope, authority, experiments, and optional continuation evidence.

- **Type:** Skill
- **Install:** `agentstack add skill-miaoy0ushan-fp-evidence-ledger`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [MiaoY0uShan](https://agentstack.voostack.com/s/miaoy0ushan)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [MiaoY0uShan](https://github.com/MiaoY0uShan)
- **Source:** https://github.com/MiaoY0uShan/FP/tree/main/install/aider/fp/evidence-ledger

## Install

```sh
agentstack add skill-miaoy0ushan-fp-evidence-ledger
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# FP: Evidence Ledger

Every completion claim needs evidence. The canonical machine record is JSON v1 validated by `contracts/evidence-ledger.v1.schema.json`; Markdown is only a human view.

## Use When

- A medium, debug, live-system, multi-agent, or risky task executed.
- A task failed, blocked, exceeded scope, or needs a durable handoff.
- Metrics or adaptive improvement will consume the result.

Small clear changes may use the Tiny Evidence form instead.

## Procedure

1. Record task ID, authority mode, write authorization, and scope.
2. List files read and touched using normalized repository-relative paths when possible.
3. Record each command/check with exact result and evidence; do not equate “ran” with “passed.”
4. For debug work, record hypothesis, probe, actual result, decision, and next step. Before completion, add `debug_evidence` with the causal chain, first divergence, condition-based wait evidence or a justified fixed wait, and distinct direct/sibling observed checks for any touched shared boundary.
5. Tie every verified claim to evidence and list every remaining unverified claim.
6. Record scope and context-budget violations explicitly.
7. Record remaining risk, decision, and next action.
8. For remote/stateful work, record baseline, rollback, effective runtime, external-client and negative-control evidence.
9. For externally retrieved context, record source, version, structured freshness basis, authoritative page, claim, and unresolved gap.
10. For incomplete cross-session work, compile a continuation block; never use it to auto-replay writes.
11. When a deliberate shortcut is accepted because it has a known safe limit, add `deferred_items` with its location, shortcut, ceiling, observable upgrade trigger, evidence, and an `evidence_ref` that resolves to observed command or verified-claim evidence. Do not create debt entries for vague future work or use them to excuse a current acceptance failure.
12. For multi-agent work, record the parent authority ceiling and every delegation envelope, including allowed resources, read-only status, summary budget, artifact path, and observed proposal evidence references.

## Completion Gate

- `result=pass` needs at least one passing check and one evidenced claim.
- `decision=complete` needs `result=pass`.
- Required checks come from the Execution Brief and must all pass.
- Report-only work cannot touch files or claim write authority.
- Unknown values stay `unknown`/`null`.
- A stale live-system observation is historical evidence, not current state.

## Output

Use `templates/evidence-ledger.md`. Run:

```text
node scripts/lint-contracts.js --ledger  --brief 
```

If the run needs metrics, pass the canonical JSON and its brief to `metrics`. Use adaptive improvement only when the ledger contains evidence for a reusable change.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [MiaoY0uShan](https://github.com/MiaoY0uShan)
- **Source:** [MiaoY0uShan/FP](https://github.com/MiaoY0uShan/FP)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-miaoy0ushan-fp-evidence-ledger
- Seller: https://agentstack.voostack.com/s/miaoy0ushan
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.