replay-buffer
/replay-buffer
Experience replay for tool/test/build failures. Invoke IMMEDIATELY when any tool fails (non-zero exit, FAIL/error/panic in output). Looks up known fixes before debugging, records failures and fixes, promotes successful fixes. Triggers on: test failures, build errors, lint failures, compilation errors, any recurring error patterns.
- Author
- zach-source ↗
- Repository
- claude-plugins ↗
- Licence
- MIT
- Last repository change
- 16.05.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 16.05.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Experience replay for tool/test/build failures. Invoke IMMEDIATELY when any tool fails (non-zero exit, FAIL/error/panic in output). Looks up known fixes before debugging, records failures and fixes, promotes successful fixes. Triggers on: test failures, build errors, lint failures, compilation errors, any recurring error patterns.
Original author description
Experience replay for tool/test/build failures. Invoke IMMEDIATELY when any tool fails (non-zero exit, FAIL/error/panic in output). Looks up known fixes before debugging, records failures and fixes, promotes successful fixes. Triggers on: test failures, build errors, lint failures, compilation errors, any recurring error patterns.
How it works
- Mandatory Workflow
- Data Management Tools
- Error Patterns to Watch
- Features
- Guardrails
SKILL.md preview
---
name: replay-buffer
description: "Experience replay for tool/test/build failures. Invoke IMMEDIATELY when any tool fails (non-zero exit, FAIL/error/panic in output). Looks up known fixes before debugging, records failures and fixes, promotes successful fixes. Triggers on: test failures, build errors, lint failures, compilation errors, any recurring error patterns."
---
# Replay Buffer
Local experience replay: lookup known fixes before debugging, record new fixes, promote on success.
## Mandatory Workflow
On ANY tool failure (non-zero exit or error output), execute in order:
### Step A: Lookup First
Before exploring fixes, call the `mcp__replay-memory__lookup_fix` tool:
```json
{
"tool": "<tool name>",
"args": ["<args>"],
"stderr": "<stderr output>",
"stdout": "<stdout if small>",
"cwd": "<repo path>"
}
```
If `confidence >= 0.7`:
- Say: "Replay buffer hit: applying prior fix: <summary>"
- Apply the fix immediately (patch or described actions)
- Skip brainstorming
If no match: proceed with normal debugging.
### Step B: Record Failure
On failure, call the `mcp__replay-memory__record_failure` tool:
```json
{
"tool": "<tool>",
"args": ["<args>"],
"exit_code": "<code>",
"stderr": "<stderr>",
"stdout": "<stdout>",
"goal": "<what I was trying to do>"
}
```
Save returned `failure_id` for correlation.
### Step C: Record Fix
After applying code edits, call the `mcp__replay-memory__record_fix` tool:
```json
{
"failure_id": "<failure_id>",
"summary": "<1-sentence specific fix, e.g. 'Mock time.Now() in TestTokenRefresh setup'>",
"patch": "<diff or file list>"
}
```
### Step D: Promote on Success
When the same tool/test now passes, call the `mcp__replay-memory__record_success` tool:
```json
{
"failure_id": "<failure_id>"
}
```
If fix didn't work: do NOT promote. Try another fix.
## Data Management Tools
Optional tools for managing replay buffer data:
### List Failures
Browse recorded failures with filters:
```json
{
"limit": 50,
"offset": 0,
"tool_filter": "go test",
"has_fix": true
}
```
### Get Statistics
View overall success rate and top failing tools:
```json
{}
```
### Export Backup
Export all failures and fixes to JSON for backup:
```json
{}
```
### Clear Old Failures
Remove failures older than N days:
```json
{
"days": 30
}
```
## Error Patterns to Watch
Invoke this skill when stderr/stdout contains:
- `FAIL` / `FAILED` / `error` / `panic` / `exception`
- `error:` / `fatal:` / `compilation terminated`
- Test framework failures
- Lint rule violations
- Build tool errors
## Features
- **Exact matching**: Normalizes line numbers, timestamps, file paths, UUIDs
- **Optional fuzzy search**: Requires `sentence-transformers` package for vector similarity
- **Secret redaction**: Automatically redacts API keys, tokens, passwords before storage
- **Confidence scoring**: Fixes promoted after 2 confirmed successes
## Guardrails
- **Redact secrets** before recording: API keys, tokens, passwords, credentials
- **Prefer minimal patches**: If diff is huge, store summary + key snippet only
- **Don't force replay** if codebase changed materially: treat as hint, not command
- **Check stats periodically**: Use `get_statistics` to see success rate and top issues
Versions
The database has no other skills named replay-buffer.
Database snapshot: 13.11.2026 · package version e6e6be84f06d (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 91analyze-failures
Analyze Playwright E2E test failure reports from CI. Parses merged blob reports, groups similar errors, and delegates to specialized subagents for investigation and fixes. Use when CI tests fail or when asked to fix E2E test failures. - 95error-explanation-generator
Explains test failures and provides actionable debugging guidance. Use when tests fail (unit, integration, E2E), builds fail, or code throws errors. Analyzes error messages, stack traces, and test output to identify root causes and suggest concrete fixes. Handles pytest, jest,… - 95runtime-error-explainer
Explains runtime errors and compilation failures with actionable debugging guidance. Use when Python or Java code throws runtime exceptions (NullPointerException, TypeError, AttributeError, etc.), compilation errors (syntax errors, type mismatches, import failures), or… - 100devtu-fix-tool
Fix failing ToolUniverse tools by diagnosing test failures, identifying root causes, implementing fixes, and validating solutions. Use when ToolUniverse tools fail tests, return errors, have schema validation issues, or when asked to debug or fix tools in the ToolUniverse framework. - 90autoresearch:fix
Iterative error-crusher loop that auto-stops at 0 errors. Cascade-aware: fixes dependency errors before their dependents. Refuses anti-patterns that hide errors instead of fixing them. TRIGGER when: user has errors or failures to fix iteratively; user asks to "fix all errors";… - 100output-error-nondeterminism
Fix non-determinism errors in Output SDK workflows. Use when seeing replay failures, inconsistent results between runs, "non-deterministic" error messages, or workflows behaving differently on retry.