Close bugs and incidents with an auditable BUG RECEIPT that states VERIFIED, PARTIAL, or BLOCKED based on baseline, root cause, change, proof, gaps, and evidence source. Use when the user asks for defect repair, regression proof, production incident closeout, issue closure, machine-readable receipt JSON, or CI integration of bug evidence.
Turn a bug fix, incident investigation, or closure decision into a strict receipt; preserve the observed defect, prove the responsible mechanism and change, then output only the receipt or its validated JSON artifact.
When to invoke
"Close this bug with proof."
"Produce a bug receipt for this incident."
"Did the fix actually verify the user-visible behavior?"
"Generate the machine-readable bug receipt JSON."
"Mark this defect VERIFIED, PARTIAL, or BLOCKED."
Prerequisites and context
Use current execution evidence when available; otherwise label evidence as supplied or mixed and do not imply it was executed now.
For JSON artifacts, read references/receipt-contract.md, start from assets/receipt.template.json, and validate with node scripts/validate-receipt.mjs <receipt.json> from this skill directory.
Do not commit a generated receipt unless the user explicitly requests it.
Evidence boundary
Before editing or closing, record the defect, intended behavior, strongest direct check, and source boundary.
Field
Required standard
Forbidden shortcut
Problem
Observed defect plus intended behavior.
A vague issue title with no observable contract.
Baseline
Failing interaction or command and decisive result, or explicit not run.
Treating an old log as a reproduction executed now.
Root cause
Proven mechanism at a concrete location or runtime transition, or explicit unproven hypothesis.
Naming a plausible component because the patch touched it.
Change
Responsible repair, or none.
Listing unrelated cleanup as proof of repair.
Proof
Executed or supplied check with every decisive layer.
Passing build, stale log, or source read alone for user-visible behavior.
Gaps
none, or the exact missing proof and one minimal next experiment/package.
Hiding gaps to make the receipt look verified.
Source
executed now, supplied, or mixed.
Blending supplied and executed evidence without labeling it.
Trace and repair rules
Follow the live owner path from input to symptom.
Separate observed facts, bounded inferences, and gaps.
Require a concrete code location, configuration setting, request transition, database transition, or runtime handoff before naming root cause.
Make the smallest responsible change.
Avoid unrelated cleanup, retries, silent fallbacks, and fixture-specific exceptions.
Do not convert a plausible patch, stale log, source read, or passing build into proof of user-visible behavior.
Direct proof boundaries
Surface
Required direct proof
Logic or failing test
Original failing input or focused regression test now passes.
UI behavior
Real interaction plus relevant console and network observation.
API or integration
Request, response, and responsible service behavior.
Persistence
Write/read or reload round trip through the real owner path.
Race or lifecycle
Repeated concurrent trigger; zero-or-one success; affected-row and transaction evidence; final invariant.
Cross-system blocker
One sanitized failing request/response with timestamp or request ID, edge and application logs, and identity-provider logs when the trace reaches that owner.
Cross-system evidence package
When the trace reaches an identity provider, include sanitized identity-provider or identity-pro evidence with the request ID, edge log, application log, and timestamp. The receipt may stay PARTIAL or BLOCKED until that correlated package is available.
Status assignment
Status
Use only when
Required gap language
VERIFIED
Baseline observed, concrete cause proven, responsible change made, all declared checks passed, and no material gap remains.
Gaps none
PARTIAL
Useful evidence exists, but a required proof layer is missing or inconclusive.
Name the single minimal experiment or correlated evidence package that closes the gap.
BLOCKED
A specific external condition prevents reproduction, repair, or proof.
Name the external blocker and the smallest evidence package needed to unblock.
Progressive disclosure and bundled resources
Resource
Use when
Contract
references/receipt-contract.md
Machine-readable receipt, CI integration, or status invariants are requested
Follow the JSON fields and status invariants exactly.
references/receipt.schema.json
Validating or explaining JSON structure
Treat as the schema source of truth.
assets/receipt.template.json
Creating a JSON artifact
Copy to a task-owned path, fill the fields, then validate.
scripts/validate-receipt.mjs
Checking a JSON receipt
Run node scripts/validate-receipt.mjs <receipt.json> from the skill directory.
Gotchas
Return the complete receipt as the entire user-facing result: concision shortens field values; it never removes or renames rows.
Use explicit placeholders: write not run, unproven, or none; do not omit a row.
Do not overclaim: supplied evidence stays supplied, and source reads are not behavioral proof.
One gap only: for PARTIAL or BLOCKED, name the decisive next experiment or evidence package, not a generic task list.
Output template
BUG RECEIPT · VERIFIED | PARTIAL | BLOCKED
Problem <observed defect and intended behavior>
Baseline <failing interaction or command and decisive result; or not run>
Root cause <proven mechanism; or unproven hypothesis>
Change <responsible change; or none>
Proof <supplied or executed check: result; include every decisive layer>
Gaps <none; or exact missing proof and single next experiment/package>
Source executed now | supplied | mixed
Quality gate
The final user-facing result is the complete BUG RECEIPT and no prose replaces it.
Status is exactly VERIFIED, PARTIAL, or BLOCKED and matches the evidence standard.
Problem, Baseline, Root cause, Change, Proof, Gaps, and Source rows are present and not renamed.
The proof includes every decisive layer for the affected surface or names the missing layer as the gap.
Supplied evidence, executed evidence, and mixed evidence are labeled honestly.
JSON artifacts, when requested, were created from assets/receipt.template.json and validated with node scripts/validate-receipt.mjs <receipt.json>.
1---2name: bug-receipt3description: Close bugs and incidents with an auditable BUG RECEIPT that states VERIFIED, PARTIAL, or BLOCKED based on baseline, root cause, change, proof, gaps, and evidence source. Use when the user asks for defect repair, regression proof, production incident closeout, issue closure, machine-readable receipt JSON, or CI integration of bug evidence.4---56<!-- Generated from harness/github-copilot/skills/bug-receipt/SKILL.md by harness/claude-code/scripts/convert_from_copilot.py. Edit the source, not this file. -->78# Bug receipt910Turn a bug fix, incident investigation, or closure decision into a strict receipt; preserve the observed defect, prove the responsible mechanism and change, then output only the receipt or its validated JSON artifact.1112## When to invoke1314- "Close this bug with proof."15- "Produce a bug receipt for this incident."16- "Did the fix actually verify the user-visible behavior?"17- "Generate the machine-readable bug receipt JSON."18- "Mark this defect VERIFIED, PARTIAL, or BLOCKED."1920## Prerequisites and context2122- Use current execution evidence when available; otherwise label evidence as `supplied` or `mixed` and do not imply it was executed now.23- For JSON artifacts, read `references/receipt-contract.md`, start from `assets/receipt.template.json`, and validate with `node scripts/validate-receipt.mjs <receipt.json>` from this skill directory.24- Do not commit a generated receipt unless the user explicitly requests it.2526## Evidence boundary2728Before editing or closing, record the defect, intended behavior, strongest direct check, and source boundary.2930| Field | Required standard | Forbidden shortcut |31| --- | --- | --- |32| Problem | Observed defect plus intended behavior. | A vague issue title with no observable contract. |33| Baseline | Failing interaction or command and decisive result, or explicit `not run`. | Treating an old log as a reproduction executed now. |34| Root cause | Proven mechanism at a concrete location or runtime transition, or explicit `unproven hypothesis`. | Naming a plausible component because the patch touched it. |35| Change | Responsible repair, or `none`. | Listing unrelated cleanup as proof of repair. |36| Proof | Executed or supplied check with every decisive layer. | Passing build, stale log, or source read alone for user-visible behavior. |37| Gaps | `none`, or the exact missing proof and one minimal next experiment/package. | Hiding gaps to make the receipt look verified. |38| Source | `executed now`, `supplied`, or `mixed`. | Blending supplied and executed evidence without labeling it. |3940## Trace and repair rules41421. Follow the live owner path from input to symptom.432. Separate observed facts, bounded inferences, and gaps.443. Require a concrete code location, configuration setting, request transition, database transition, or runtime handoff before naming root cause.454. Make the smallest responsible change.465. Avoid unrelated cleanup, retries, silent fallbacks, and fixture-specific exceptions.476. Do not convert a plausible patch, stale log, source read, or passing build into proof of user-visible behavior.4849## Direct proof boundaries5051| Surface | Required direct proof |52| --- | --- |53| Logic or failing test | Original failing input or focused regression test now passes. |54| UI behavior | Real interaction plus relevant console and network observation. |55| API or integration | Request, response, and responsible service behavior. |56| Persistence | Write/read or reload round trip through the real owner path. |57| Race or lifecycle | Repeated concurrent trigger; zero-or-one success; affected-row and transaction evidence; final invariant. |58| Cross-system blocker | One sanitized failing request/response with timestamp or request ID, edge and application logs, and identity-provider logs when the trace reaches that owner. |5960## Cross-system evidence package6162When the trace reaches an identity provider, include sanitized identity-provider or `identity-pro` evidence with the request ID, edge log, application log, and timestamp. The receipt may stay `PARTIAL` or `BLOCKED` until that correlated package is available.63## Status assignment6465| Status | Use only when | Required gap language |66| --- | --- | --- |67| `VERIFIED` | Baseline observed, concrete cause proven, responsible change made, all declared checks passed, and no material gap remains. | `Gaps none` |68| `PARTIAL` | Useful evidence exists, but a required proof layer is missing or inconclusive. | Name the single minimal experiment or correlated evidence package that closes the gap. |69| `BLOCKED` | A specific external condition prevents reproduction, repair, or proof. | Name the external blocker and the smallest evidence package needed to unblock. |7071## Progressive disclosure and bundled resources7273| Resource | Use when | Contract |74| --- | --- | --- |75| `references/receipt-contract.md` | Machine-readable receipt, CI integration, or status invariants are requested | Follow the JSON fields and status invariants exactly. |76| `references/receipt.schema.json` | Validating or explaining JSON structure | Treat as the schema source of truth. |77| `assets/receipt.template.json` | Creating a JSON artifact | Copy to a task-owned path, fill the fields, then validate. |78| `scripts/validate-receipt.mjs` | Checking a JSON receipt | Run `node scripts/validate-receipt.mjs <receipt.json>` from the skill directory. |7980## Gotchas8182- **Return the complete receipt as the entire user-facing result**: concision shortens field values; it never removes or renames rows.83- **Use explicit placeholders**: write `not run`, `unproven`, or `none`; do not omit a row.84- **Do not overclaim**: supplied evidence stays supplied, and source reads are not behavioral proof.85- **One gap only**: for `PARTIAL` or `BLOCKED`, name the decisive next experiment or evidence package, not a generic task list.8687## Output template8889```markdown90BUG RECEIPT · VERIFIED | PARTIAL | BLOCKED9192Problem <observed defect and intended behavior>93Baseline <failing interaction or command and decisive result; or not run>94Root cause <proven mechanism; or unproven hypothesis>95Change <responsible change; or none>96Proof <supplied or executed check: result; include every decisive layer>97Gaps <none; or exact missing proof and single next experiment/package>98Source executed now | supplied | mixed99```100101## Quality gate102103- [ ] The final user-facing result is the complete BUG RECEIPT and no prose replaces it.104- [ ] Status is exactly `VERIFIED`, `PARTIAL`, or `BLOCKED` and matches the evidence standard.105- [ ] `Problem`, `Baseline`, `Root cause`, `Change`, `Proof`, `Gaps`, and `Source` rows are present and not renamed.106- [ ] The proof includes every decisive layer for the affected surface or names the missing layer as the gap.107- [ ] Supplied evidence, executed evidence, and mixed evidence are labeled honestly.108- [ ] JSON artifacts, when requested, were created from `assets/receipt.template.json` and validated with `node scripts/validate-receipt.mjs <receipt.json>`.109110## References111112- [Original bug-receipt repository](https://github.com/lMysticl/bug-receipt)
Run npx skillmds@latest add paulasilvatech/bug-receipt 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.
Close bugs and incidents with an auditable BUG RECEIPT that states VERIFIED, PARTIAL, or BLOCKED based on baseline, root cause, change, proof, gaps, and evidence source. Use when the user asks for defect repair, regression proof, production incident closeout, issue closure, machine-readable receipt JSON, or CI integration of bug evidence. It is listed under Integrations & APIs on SkillMD.
This skill has not completed SkillMD's automated safety review yet. 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.
paulasilvatech (@paulasilvatech) published this skill. Their other Agent Skills are listed on their SkillMD profile.
paulasilvatech