agenttrace-session-audit

Name: agenttrace-session-audit
Author: sickn33

44kstars6.5kforks

How to install agenttrace-session-audit?

sickn33/agentic-awesome-skills/agenttrace-session-audit

$npx -y skills add sickn33/agentic-awesome-skills --skill agenttrace-session-audit

Installs into the current project.

Summary

Audit local AI coding-agent sessions with agenttrace for cost, tool failures, latency, anomalies, health, diffs, and CI gates.

Files · 1

View on GitHub

SKILL.md

1	# agenttrace Session Audit
2
3	## Overview
4
5	Use this skill to inspect local AI coding-agent sessions with
6	[agenttrace](https://github.com/luoyuctl/agenttrace). It focuses on the process
7	behind a run: token and cost spikes, tool failures, retry loops, latency gaps,
8	anomalies, health scores, and session-to-session diffs.
9
10	agenttrace is local-first and reads session logs from tools such as Claude Code,
11	Codex CLI, Gemini CLI, Aider, Cursor exports, OpenCode, Qwen Code, Kimi, and
12	generic JSON or JSONL traces.
13
14	## When to Use This Skill
15
16	- Use when a user asks why an AI coding run was slow, expensive, shallow, or unreliable.
17	- Use when reviewing local agent logs before retrying a failed or suspicious task.
18	- Use when building a lightweight CI health gate for AI-assisted coding sessions.
19	- Use when comparing two attempts and looking for changed tool paths, retries, or cost patterns.
20
21	## How It Works
22
23	### Step 1: Discover Available Sessions
24
25	Prefer an installed `agenttrace` binary when it is available on `PATH`. If the
26	current repository is `luoyuctl/agenttrace`, use `go run ./cmd/agenttrace`
27	instead.
28
29	```bash
30	agenttrace --doctor
31	agenttrace --overview
32	```
33
34	If no sessions are detected, report the directories checked by `--doctor` and
35	ask for the exported session file or log directory.
36
37	### Step 2: Produce a Human-Readable Audit
38
39	Use Markdown when the user wants a concise report they can inspect or share.
40
41	```bash
42	agenttrace --overview -f markdown -o agenttrace-overview.md
43	```
44
45	In the report, lead with the highest-risk sessions and explain why they matter:
46	critical anomalies, repeated tool failures, token or cost waste, long latency
47	gaps, low health scores, and suspiciously shallow sessions.
48
49	### Step 3: Inspect One Session or Directory
50
51	Use the latest session for a quick check, or pass an explicit export path when
52	the user provides one.
53
54	```bash
55	agenttrace --latest
56	agenttrace --latest -f json
57	agenttrace path/to/session-or-export.json
58	agenttrace --overview -d path/to/session-dir
59	```
60
61	### Step 4: Compare Attempts When Semantics Matter
62
63	Token and latency metrics can look healthy even when an agent confidently takes
64	the wrong implementation path. When the risk is semantic drift, pair the trace
65	audit with a diff against a previous or known-good attempt.
66
67	Look for:
68
69	- changed files or commands that diverge from the intended task
70	- missing tests or verification steps compared with the reference attempt
71	- repeated edits around the same files without a clear reason
72	- lower cost that came from skipping necessary exploration
73
74	### Step 5: Add Automation Gates
75
76	For CI or repeatable team workflows, use JSON output or health thresholds.
77
78	```bash
79	agenttrace --overview -f json -o agenttrace-overview.json
80	agenttrace --overview --fail-under-health 80 --fail-on-critical --max-tool-fail-rate 15
81	```
82
83	Tune thresholds to the project. A strict gate is useful for critical workflows;
84	a reporting-only command is better while the team is learning its baseline.
85
86	## Examples
87
88	### Quick Local Review
89
90	```bash
91	agenttrace --overview
92	agenttrace --latest
93	```
94
95	Use this after a long coding-agent run to decide whether the next prompt should
96	split the task, avoid a failing tool path, add missing tests, or reset context.
97
98	### CI Health Check
99
100	```bash
101	agenttrace --overview --fail-under-health 80 --fail-on-critical
102	```
103
104	Use this when agent session logs are available in CI and the team wants a simple
105	guard against critical anomalies or unhealthy runs.
106
107	## Best Practices
108
109	- Start with `--doctor` when session discovery is uncertain.
110	- Report missing fields plainly; do not invent cost, model, latency, or health data.
111	- Treat prompts, code, and session contents as private local data.
112	- Prefer JSON output for automation and Markdown output for human review.
113	- Use trace metrics for process failures and diff/reference review for semantic drift.
114
115	## Limitations
116
117	- agenttrace can only analyze logs that are present locally or provided as exports.
118	- Some agents do not expose enough fields to infer cost, model, cache use, or latency.
119	- Healthy trace metrics do not prove the final code is correct; still run tests and review diffs.
120	- CI gates should start as advisory until the team understands normal baseline behavior.
121
122	## Security & Safety Notes
123
124	- Do not upload private session logs to external services unless the user explicitly approves it.
125	- Do not overwrite user reports unless they requested that exact output path.
126	- Avoid printing secrets found in p

People who liked this also viewed

6 picks

Type

entra-app-registrationGuides Microsoft Entra ID app registration, OAuth 2.0 authentication, and MSAL integration. USE FOR: create app registration, register Azure AD app, configure OAuth, set up authentication, add API perSkillsJul 2026471k1.3k
azure-complianceRun Azure compliance and security audits with azqr plus Key Vault expiration checks. Covers best-practice assessment, resource review, policy/compliance validation, and security posture checks. WHEN: SkillsJul 2026470k1.3k
azure-rbacOfficial agent plugin providing skills and MCP server configurations for Azure scenarios.SkillsJul 2026466k1.3k
better-auth-best-practicesAgent skill from better-auth/skillsSkillsJul 202678k203
firebase-security-rules-auditorA skill to evaluate how secure Firestore security rules are. Use this when Firestore security rules are updated to ensure that the generated rules are extremely secure and robust.SkillsJul 202676k386
audit-websiteAudit a website with the squirrelscan CLI and fix the findings in code. Runs SEO, performance, security, technical, content, accessibility, and 15 other rule categories (260+ rules), returns an LLM-opSkillsJul 202665k85

agenttrace-session-audit

44kstars6.5kforks

View on GitHub

How to install agenttrace-session-audit?

sickn33/agentic-awesome-skills/agenttrace-session-audit

$npx -y skills add sickn33/agentic-awesome-skills --skill agenttrace-session-audit

Installs into the current project.

Summary

Audit local AI coding-agent sessions with agenttrace for cost, tool failures, latency, anomalies, health, diffs, and CI gates.

Files · 1

View on GitHub

SKILL.md

1	# agenttrace Session Audit
2
3	## Overview
4
5	Use this skill to inspect local AI coding-agent sessions with
6	[agenttrace](https://github.com/luoyuctl/agenttrace). It focuses on the process
7	behind a run: token and cost spikes, tool failures, retry loops, latency gaps,
8	anomalies, health scores, and session-to-session diffs.
9
10	agenttrace is local-first and reads session logs from tools such as Claude Code,
11	Codex CLI, Gemini CLI, Aider, Cursor exports, OpenCode, Qwen Code, Kimi, and
12	generic JSON or JSONL traces.
13
14	## When to Use This Skill
15
16	- Use when a user asks why an AI coding run was slow, expensive, shallow, or unreliable.
17	- Use when reviewing local agent logs before retrying a failed or suspicious task.
18	- Use when building a lightweight CI health gate for AI-assisted coding sessions.
19	- Use when comparing two attempts and looking for changed tool paths, retries, or cost patterns.
20
21	## How It Works
22
23	### Step 1: Discover Available Sessions
24
25	Prefer an installed `agenttrace` binary when it is available on `PATH`. If the
26	current repository is `luoyuctl/agenttrace`, use `go run ./cmd/agenttrace`
27	instead.
28
29	```bash
30	agenttrace --doctor
31	agenttrace --overview
32	```
33
34	If no sessions are detected, report the directories checked by `--doctor` and
35	ask for the exported session file or log directory.
36
37	### Step 2: Produce a Human-Readable Audit
38
39	Use Markdown when the user wants a concise report they can inspect or share.
40
41	```bash
42	agenttrace --overview -f markdown -o agenttrace-overview.md
43	```
44
45	In the report, lead with the highest-risk sessions and explain why they matter:
46	critical anomalies, repeated tool failures, token or cost waste, long latency
47	gaps, low health scores, and suspiciously shallow sessions.
48
49	### Step 3: Inspect One Session or Directory
50
51	Use the latest session for a quick check, or pass an explicit export path when
52	the user provides one.
53
54	```bash
55	agenttrace --latest
56	agenttrace --latest -f json
57	agenttrace path/to/session-or-export.json
58	agenttrace --overview -d path/to/session-dir
59	```
60
61	### Step 4: Compare Attempts When Semantics Matter
62
63	Token and latency metrics can look healthy even when an agent confidently takes
64	the wrong implementation path. When the risk is semantic drift, pair the trace
65	audit with a diff against a previous or known-good attempt.
66
67	Look for:
68
69	- changed files or commands that diverge from the intended task
70	- missing tests or verification steps compared with the reference attempt
71	- repeated edits around the same files without a clear reason
72	- lower cost that came from skipping necessary exploration
73
74	### Step 5: Add Automation Gates
75
76	For CI or repeatable team workflows, use JSON output or health thresholds.
77
78	```bash
79	agenttrace --overview -f json -o agenttrace-overview.json
80	agenttrace --overview --fail-under-health 80 --fail-on-critical --max-tool-fail-rate 15
81	```
82
83	Tune thresholds to the project. A strict gate is useful for critical workflows;
84	a reporting-only command is better while the team is learning its baseline.
85
86	## Examples
87
88	### Quick Local Review
89
90	```bash
91	agenttrace --overview
92	agenttrace --latest
93	```
94
95	Use this after a long coding-agent run to decide whether the next prompt should
96	split the task, avoid a failing tool path, add missing tests, or reset context.
97
98	### CI Health Check
99
100	```bash
101	agenttrace --overview --fail-under-health 80 --fail-on-critical
102	```
103
104	Use this when agent session logs are available in CI and the team wants a simple
105	guard against critical anomalies or unhealthy runs.
106
107	## Best Practices
108
109	- Start with `--doctor` when session discovery is uncertain.
110	- Report missing fields plainly; do not invent cost, model, latency, or health data.
111	- Treat prompts, code, and session contents as private local data.
112	- Prefer JSON output for automation and Markdown output for human review.
113	- Use trace metrics for process failures and diff/reference review for semantic drift.
114
115	## Limitations
116
117	- agenttrace can only analyze logs that are present locally or provided as exports.
118	- Some agents do not expose enough fields to infer cost, model, cache use, or latency.
119	- Healthy trace metrics do not prove the final code is correct; still run tests and review diffs.
120	- CI gates should start as advisory until the team understands normal baseline behavior.
121
122	## Security & Safety Notes
123
124	- Do not upload private session logs to external services unless the user explicitly approves it.
125	- Do not overwrite user reports unless they requested that exact output path.
126	- Avoid printing secrets found in p

People who liked this also viewed

6 picks

Type