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.
npx agentmods add skills/gg-mo/second-thought-agent-skill/second-thoughtnpx skills add gg-mo/second-thought-agent-skill --skill second-thoughtgit clone --depth 1 https://github.com/gg-mo/second-thought-agent-skillWrote 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/skills/gg-mo/second-thought-agent-skill/second-thought)<a href="https://agentmods.dev/skills/gg-mo/second-thought-agent-skill/second-thought"><img src="https://agentmods.dev/badge/skills/gg-mo/second-thought-agent-skill/second-thought.svg" alt="Measured on agentmods" 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 | $0.00049 | $0.00691 |
| Opus 5 | $0.00024 | $0.00345 |
| Sonnet 5 | $0.00010 | $0.00138 |
| Haiku 4.5 | $0.00005 | $0.00069 |
Grade A, and why
second-thought 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 4d 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Second Thought
A standalone decision checkpoint engine that asks whether the current direction is actually right before the agent commits.
Core rule
Before moving forward with a meaningful proposal, run a critique against the original intent, not just the latest turn.
Default behavior: auto-build the payload from conversation context. Do not require manual field entry unless the user explicitly wants manual/debug mode.
Input contract
Provide these fields to the critique engine:
original_intentcurrent_contextproposal_type(idea | decision | design | plan | action | tool_call | response)proposalrationaleconstraintsrisk_level(low | medium | high)stage(brainstorming | convergence | pre_execution | post_feedback)should_challenge
Output contract
Produce this structured result internally:
status(proceed | revise | prompt_human)summarygoal_alignmentconcernsassumptionsbetter_optionschallenge_promptrecommended_next_stepprompt_to_human
User-facing response contract
Default output should be human-readable, not raw JSON.
- Use a conversational response in a few sentences (about 2-4).
- Avoid rigid section labels like
Decision:,Why:,Next:unless explicitly requested. - Include the decision (
proceed | revise | prompt_human) naturally in prose. - Include at least one specific reason grounded in the current proposal/context.
- Include one concrete next step.
- If
prompt_human, ask exactly one focused clarification question.
Only show full JSON when the human explicitly requests JSON/debug/contract format.
Stage behavior
brainstorming: exploratory but critical; do not reward weak framingconvergence: skeptical; test if agreement is correctness vs momentumpre_execution: strict; verify reversibility, risk, missing informationpost_feedback: re-evaluate without overfitting to latest feedback
Decision standard
proceed: strong intent alignment, acceptable assumptions, no required human clarificationrevise: direction is promising but flawed, drifting, or weakly justifiedprompt_human: ambiguity, under-specification, high impact, irreversibility, or clear preference/judgment call needed
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.
- 4d ago First seen · 98 lines · 49 tokens per session scan A 09b914829abf
second-thought is a skill published in the GitHub repository gg-mo/second-thought-agent-skill (5 stars, last pushed 4mo ago), licensed MIT. It adds 49 tokens to every session and 691 once invoked, about $0.0002 per session on Opus 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 2026-08-31.
Other skills, from other repositories
create-diagram
Create architecture diagrams from a local codebase or system description. Use when the user asks to diagram a repo, map system architecture, show component relationships, create drill-down views, or explain a system visually with tld.
b2-cloud-storage
Manage Backblaze B2 cloud storage. List files, audit usage, estimate cost, clean up stale data, review security posture, and manage lifecycle rules. Use when the user mentions B2, Backblaze, object storage buckets, or storage cleanup.
application-builder
Build a targeted resume and cover letter for a role folder that already has a job description and role analysis. Runs iterative adversarial review rounds until the materials stabilize, then hands off for human review before submission.
outreach
Open the warm channel for a role the pipeline already assessed. Finds who the user already knows at the company, researches named targets from public professional presence when they know nobody, grades each path honestly, drafts the ask through the voice agent, and writes the contacts into the Referrals table in…
adversarial-review
Run a seven-agent adversarial pipeline against a resume and cover letter before submission. Produces channel-conditional verdicts anchored to the realistic applicant pool, separates materials quality from submission expected value, and ends with a contrarian pass that can overturn the verdict.
redesign-templates
Redesign the resume and cover letter templates into your own house style. Changes typography, palette, and layout while holding the ATS constraints fixed, then proves the result still compiles and still extracts. Run any time; not a pipeline stage.