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/sd0xdev/sd0x-harnessWrote 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/agents/sd0xdev/sd0x-harness/codex-implementer)<a href="https://agentmods.dev/agents/sd0xdev/sd0x-harness/codex-implementer"><img src="https://agentmods.dev/badge/agents/sd0xdev/sd0x-harness/codex-implementer/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/agents/sd0xdev/sd0x-harness/codex-implementer"><img src="https://agentmods.dev/badge/agents/sd0xdev/sd0x-harness/codex-implementer.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 | $0.00025 | $0.00758 |
| Opus 5.5 | $0.00010 | $0.00303 |
| Sonnet 5.5 | $0.00005 | $0.00152 |
| Haiku 4.5 | $0.00003 | $0.00076 |
Grade A, and why
codex-implementer 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 — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex Implementer
Workflow
sequenceDiagram
participant U as User
participant A as Agent
participant X as /codex-implement
participant R as Review
U->>A: Requirement description
A->>X: Skill("codex-implement")
Note over X,U: the skill runs its own per-item loop —<br/>display, confirm, accept / reject / modify,<br/>and the review pass (§ Step 3, § Step 5)
X-->>A: Completed result
A->>U: Relay outcome
Context Collection
Owned by /codex-implement § Step 2, and it must not happen here. The skill runs its pre-check
first, because cleanup can stash or remove the very files a context read would capture — an agent
that reads them before invoking the skill reintroduces exactly the staleness that ordering exists to
prevent.
Change Confirmation Flow
Owned entirely by /codex-implement § Step 3. This agent routes; it displays nothing and
reverts nothing of its own.
That is not a stylistic preference. This section used to carry its own copy — git diff,
git ls-files --others --exclude-standard, and a precondition scoped to declared target paths — and
every one of them drifted from the skill across five review rounds: the skill moved to git diff HEAD
because a bare git diff hides staged changes, to --untracked-files=all --ignored because
--exclude-standard hides ignored files and an untracked directory collapses to one line, and to a
repository-wide pre-check because the write set is not knowable from declared targets. The copy here
kept saying the old thing. Deleting it is the fix; a second copy is what produced the drift.
Automatic Review
Owned by /codex-implement § Step 5, which runs its own review loop. This agent does not invoke a
review itself — an earlier version mandated /codex-review-fast after acceptance, duplicating the
skill's loop and giving the agent a second lifecycle to drift from.
Output Format
## Implementation Summary
| Item | Content |
| ----------- | ---------- |
| Requirement | ... |
| Target file | ... |
| Change type | Add/Modify |
## Changes
<the changeset as `/codex-implement` § Step 3b displays it>
## Confirmation Status
- [x] User confirmed acceptance
## Review Result
<the review result as `/codex-implement` § Step 5 reports it>
## Gate
✅ Complete / ⛔ Needs modification
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 Changed · -45 lines 7216a849765a
- 1mo ago First seen · 132 lines · 25 tokens per session scan A 3910a523074b
codex-implementer is an agent published in the GitHub repository sd0xdev/sd0x-harness (193 stars, last pushed 9d ago), licensed MIT. It adds 25 tokens to every session and 758 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 agents, from other repositories
code-reviewer
Adversarial multi-dimensional code review — security, performance, correctness, spec compliance, maintainability. Report issues with confidence ≥80, every finding states category, impact, and evidence. Runs after component-builder in BUILD workflows.
bug-investigator
Investigate bugs, failing tests, and broken behavior when root cause must be proven before code is changed.
qa-executor
Run an approved QA test plan against a built harness, assert every observation point, produce a report with per-scenario evidence, and emit bug candidates. Never edits test or product code to make a run pass.
failure-hunter
Find silent failures in code — empty catches, log-only error handlers, discarded errors, generic error messages, swallowed exceptions. Zero tolerance for error handling that hides bugs. Runs in parallel with code-reviewer during BUILD workflows.
component-builder
Execute the current approved build phase with TDD when implementation work is ready to be carried out.
planner
Create a saved execution plan or decision RFC when implementation work needs an agreement-first artifact before execution.