Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/korovin-aa97/talkthrough-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/korovin-aa97/talkthrough-mcp/bug)<a href="https://agentmods.dev/commands/korovin-aa97/talkthrough-mcp/bug"><img src="https://agentmods.dev/badge/commands/korovin-aa97/talkthrough-mcp/bug/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/korovin-aa97/talkthrough-mcp/bug"><img src="https://agentmods.dev/badge/commands/korovin-aa97/talkthrough-mcp/bug.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00031 | $0.00820 |
| Opus 5.5 | $0.00012 | $0.00328 |
| Sonnet 5.5 | $0.00006 | $0.00164 |
| Haiku 4.5 | $0.00003 | $0.00082 |
Grade A, and why
bug scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
If no job_id was passed as an argument: call list_jobs() to find the right recording, or process_media(path) if the user gave a file path — then follow the workflow below with that job_id.
You are turning a recorded bug into ONE evidence-backed GitHub issue draft.
The recording was processed by talkthrough as job $ARGUMENTS. Evidence comes
before everything else — and this is a bug report, not a fix: change no code.
Method — evidence first
- Orient. Short recording: get_transcript(job_id="$ARGUMENTS") and read it whole. Long recording: search(job_id="$ARGUMENTS", query="") (error text, feature names) and read only the relevant ranges. If the job_id looks wrong, verify with list_jobs().
- A silent recording (no narration) is a VALID input, not an error: get_transcript returns an empty transcript with a note (and list_jobs shows has_transcript: false), while frames and on-screen text are still indexed — orient with search (OCR hits) and get_frames across the timeline instead.
- Pick ONE bug — the highest-confidence, highest-severity problem the evidence supports. Mention anything else in a single "Also observed" line at the end of the draft; do not investigate it.
- Evidence bundle: get_moment(job_id="$ARGUMENTS", start_ms=, end_ms=<t1+2000>) around the key remark or on-screen failure. Inspect at least one returned frame with your own eyes. Describe the observed state from the pixels and OCR text — never from imagination.
- Small text unreadable (error codes, request IDs, on-screen log lines)? extract_frame(job_id="$ARGUMENTS", at_ms=, crop=) for a native-resolution look before quoting it.
Checkpoint — write this block before drafting
- Heard: the narrator's exact words + timestamp (or "silent recording").
- Saw: the concrete visible state + OCR identifiers (error text, codes, IDs).
- When: t_wall + its confidence (or "relative timestamp only").
- Expected: the expected behavior, as stated or implied by the recording.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 66 lines · 31 tokens per session scan A 80b679ce5a02
bug is a command published in the GitHub repository korovin-aa97/talkthrough-mcp (32 stars, last pushed today), licensed MIT. It adds 31 tokens to every session and 820 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
preflight
Diagnostique l'environnement Cortex et guide la réparation (DB, extensions, modèles).
umt-doctor
Diagnose UMT MCP server setup — builds, environment variables, and config health.
ultra-agents
Autonomous multi-round review + fix loop — no per-round approval.
api-aqa-flow-execution-and-report-analysis
Phase 6 Execution & Report Analysis of api-aqa-flow (USER INTERACTION REQUIRED).
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.