一键在 Manus 中运行任何 Skill

run-trace

星标3

分支0

更新时间2026年7月5日 07:23

Append structured execution traces across operational, cognitive, and contextual surfaces with minimal overhead. Load when inspecting agent runs, logging tool calls and observations, enabling post-run debugging, or pairing with structured-planning step IDs. Also triggers on "trace this run", "log execution", "agent observability", "run log", or when fault-localize needs evidence. Default-on during multi-step plans. Traces live at .agent-loom/traces/ — git-ignored by default.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

dvy1987

dvy1987/agent-loom

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

Run Trace

You append low-overhead, structured records per meaningful step — tool call, observation, reflection, error. Three surfaces: operational, cognitive, contextual. Never block execution to log.

Hard Rules

Append to .agent-loom/traces/<run-id>.jsonl — one JSON object per line per references/TRACE-SCHEMA.md. Reuse step_id from active structured-planning plan when present. Never persist secrets — refs only in input_ref / output_ref. Logging is append-only — never rewrite prior lines. Ensure .agent-loom/traces/ is gitignored in consumer projects.

Workflow

Step 1 — Allocate run_id

Format: YYYY-MM-DDTHH:MM:SSZ-<slug> at run start. Create empty jsonl file.

Step 2 — Append per step

After each meaningful event, append one record with correct surface:

Event	Surface
Tool/command executed	operational
Plan/route/reflection	cognitive
cwd, branch, versions	contextual

Step 3 — On error

Set error field on the operational record. Continue tracing if run continues.

Step 4 — Query when needed

python3 .agents/skills/run-trace/scripts/trace_query.py <path> timeline
python3 .agents/skills/run-trace/scripts/trace_query.py <path> errors

Step 5 — Hand off

On failed run, pass trace path to fault-localize.

Gotchas

Pretty-printed JSON breaks JSONL — single-line objects only.
Logging raw env vars may leak secrets — whitelist keys only.
High-frequency polling loops — batch or sample; don't log every poll.

Output Format

## Run trace — [run_id]

Trace file: `.agent-loom/traces/[run_id].jsonl`
Records: N | Errors: N

Latest:
- [ts] [surface] [step_id] [action]

Query: trace_query.py [path] timeline

Examples

Teaser: 5-step plan run → 12 operational + 3 cognitive records → S3 error captured with exit code.

Full pairs: references/examples.md

Common Rationalizations

Excuse	Reality
"Tracing is too heavy"	One JSON line per step is cheap.
"I'll remember what happened"	You won't — especially after revert.
"Chat history is enough"	Not structured; can't query errors programmatically.
"Secrets in output are fine locally"	Traces get committed, shared, uploaded — refs only.
"Skip cognitive surface"	Route decisions are the hardest to debug.

Verification

Trace file exists and is valid JSONL
step_ids align with plan when planning active
No secrets in trace payloads
traces/ gitignored

Red Flags

Trace file with invalid JSON lines
Missing error field on failed tool steps
Secrets written to jsonl

Prune Log

Last pruned: 2026-07-05

Initial release from high-leverage skill spec (Skill 4 family)

Impact Report

Trace: [run_id] | Records: N | Errors: N | Path: .agent-loom/traces/...

同仓库更多 Skills

同仓库

code-review-crsp

dvy1987/agent-loom

Review code changes for correctness, completeness, bugs, edge cases, and quality. Load when the user explicitly asks to review code, check a PR, review a diff, audit recent changes, or verify an implementation matches requirements. Also triggers on "review this code", "check this PR", "review my changes", "code review", "did this implement correctly", "audit this diff", or any explicit request for a formal code review. Do NOT load for "review changes for context" or "review what happened" — those are requests to read code, not to perform a formal review.

2026-07-093

debug-and-fix

dvy1987/agent-loom

Fix broken or failing functionality through structured reproduction, root-cause analysis, minimal fix, and verification. Load when the user asks to fix a bug, debug an error, resolve an issue, or work on a Linear ticket. Also triggers on "this is broken", "fix this bug", "why is this failing", "debug this", "resolve this error", "what went wrong", or any request to diagnose and fix a problem.

2026-07-093

git-workflow-and-versioning

dvy1987/agent-loom

Structure git work with atomic commits, conventional messages, short-lived branches, and pre-commit hygiene. Load when committing, branching, resolving merge conflicts, organizing parallel work, or reviewing git history. Also triggers on "git workflow", "commit message", "conventional commits", "atomic commit", "branch strategy", "git worktree", "how should I commit this". Does not replace project-specific hook policies — complements them.

2026-07-093

implementation-plan

dvy1987/agent-loom

Create a detailed, step-by-step implementation plan for a feature or project. Load when the user asks to plan a feature, create a technical roadmap, break down a PRD into tasks, design an implementation strategy, or sequence engineering work. Also triggers on "how should we build this", "implementation plan for", "technical breakdown", "task list for", or any request to turn a high-level requirement into a concrete execution plan. Supports phased rollouts, architecture-first, and MVP-focused planning.

2026-07-093

incremental-implementation

dvy1987/agent-loom

Deliver multi-file changes in thin vertical slices — implement, test, verify, commit, repeat. Load when implementing a feature from a plan, building across more than one file, refactoring, or when tempted to land a large change in one pass. Also triggers on "incremental implementation", "vertical slice", "thin slice", "one slice at a time", "don't do it all at once". Routes single-file fixes to direct execution. Pairs with test-driven-development and git-workflow-and-versioning.

2026-07-093

test-driven-development

dvy1987/agent-loom

Apply the Red-Green-Refactor cycle to software development. Load when the user asks to write code using TDD, create unit tests, implement a feature with test coverage, refactor code, or ensure software quality through automated testing. Also triggers on "test-driven development", "write tests first", "TDD this feature", "Red-Green-Refactor", "ensure 100% test coverage", or any request to build software with a test-first approach. Supports unit, integration, and end-to-end testing strategies.

2026-07-093

name	run-trace
description	Append structured execution traces across operational, cognitive, and contextual surfaces with minimal overhead. Load when inspecting agent runs, logging tool calls and observations, enabling post-run debugging, or pairing with structured-planning step IDs. Also triggers on "trace this run", "log execution", "agent observability", "run log", or when fault-localize needs evidence. Default-on during multi-step plans. Traces live at .agent-loom/traces/ — git-ignored by default.
license	MIT
metadata	{"author":"dvy1987","version":"1.0","category":"project-specific","sources":"AgentTrace arXiv 2602.10133, Watson arXiv 2411.03455","resources":{"references":["TRACE-SCHEMA.md","examples.md"],"scripts":["trace_query.py"]}}

Run Trace

You append low-overhead, structured records per meaningful step — tool call, observation, reflection, error. Three surfaces: operational, cognitive, contextual. Never block execution to log.

Hard Rules

Workflow

Step 1 — Allocate run_id

Format: YYYY-MM-DDTHH:MM:SSZ-<slug> at run start. Create empty jsonl file.

Step 2 — Append per step

After each meaningful event, append one record with correct surface:

Event	Surface
Tool/command executed	operational
Plan/route/reflection	cognitive
cwd, branch, versions	contextual

Step 3 — On error

Set error field on the operational record. Continue tracing if run continues.

Step 4 — Query when needed

python3 .agents/skills/run-trace/scripts/trace_query.py <path> timeline
python3 .agents/skills/run-trace/scripts/trace_query.py <path> errors

Step 5 — Hand off

On failed run, pass trace path to fault-localize.

Gotchas

Pretty-printed JSON breaks JSONL — single-line objects only.
Logging raw env vars may leak secrets — whitelist keys only.
High-frequency polling loops — batch or sample; don't log every poll.

Output Format

## Run trace — [run_id]

Trace file: `.agent-loom/traces/[run_id].jsonl`
Records: N | Errors: N

Latest:
- [ts] [surface] [step_id] [action]

Query: trace_query.py [path] timeline

Examples

Teaser: 5-step plan run → 12 operational + 3 cognitive records → S3 error captured with exit code.

Full pairs: references/examples.md

Common Rationalizations

Excuse	Reality
"Tracing is too heavy"	One JSON line per step is cheap.
"I'll remember what happened"	You won't — especially after revert.
"Chat history is enough"	Not structured; can't query errors programmatically.
"Secrets in output are fine locally"	Traces get committed, shared, uploaded — refs only.
"Skip cognitive surface"	Route decisions are the hardest to debug.

Verification

Trace file exists and is valid JSONL
step_ids align with plan when planning active
No secrets in trace payloads
traces/ gitignored

Red Flags

Trace file with invalid JSON lines
Missing error field on failed tool steps
Secrets written to jsonl

Prune Log

Last pruned: 2026-07-05

Initial release from high-leverage skill spec (Skill 4 family)

Impact Report

Trace: [run_id] | Records: N | Errors: N | Path: .agent-loom/traces/...