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 skills add WhiteBite/opencode-dejavu --skill dejavugit clone --depth 1 https://github.com/WhiteBite/opencode-dejavuWrote 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/whitebite/opencode-dejavu/dejavu)<a href="https://agentmods.dev/skills/whitebite/opencode-dejavu/dejavu"><img src="https://agentmods.dev/badge/skills/whitebite/opencode-dejavu/dejavu/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/skills/whitebite/opencode-dejavu/dejavu"><img src="https://agentmods.dev/badge/skills/whitebite/opencode-dejavu/dejavu.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 | $0.00099 | $0.01122 |
| Opus 5.5 | $0.00040 | $0.00449 |
| Sonnet 5.5 | $0.00020 | $0.00224 |
| Haiku 4.5 | $0.00010 | $0.00112 |
Grade A, and why
dejavu 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 — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.
dejavu — recurring-error gates
dejavu is an OpenCode plugin that watches tool calls fail, counts recurrences across sessions, and promotes frequent failures into gates. It is your external long-term memory for mistakes: what failed before will not silently fail again.
You do not manage dejavu's detection — it is mechanical. Your job is to react correctly and to improve gate quality when you learn something.
When you get [dejavu] REMINDER
The call you were about to make has failed repeatedly in the past. The call was aborted before execution.
- Do not retry the identical call. That is exactly the behavior the gate exists to prevent.
- Read
Last failure:andCorrection:in the message. - Diagnose the root cause (read the relevant file, check the environment, run a diagnostic command).
- Retry with a changed approach — a different command, fixed arguments, or a prerequisite step first.
- If you are confident the situation changed (e.g. you just installed the missing dependency), retry as-is — a SUCCESS clears your session from the gate's chain. On a blocking gate, a repeated failure after a reminder escalates it to a hard block within this session; a reminding gate (diagnostics/iteration commands) never blocks, it only reminds.
dejavu:proceed(as a trailing COMMENT inside the call —# dejavu:proceed) bypasses the gate. Use it ONLY when the user explicitly asked you to force the operation, or you have concrete proof the gate is stale. Every override is logged; on blocking gates it is ALSO counted on the gate, and repeated overrides demote the gate (your bypasses are feedback: a gate everyone works around retires itself). If you override and the call SUCCEEDS, say so when improving the gate — that proves it stale.
When you get [dejavu] BLOCKED
You were reminded, retried, and it failed again. The gate is now hard in this session.
- Do not attempt the same call again in any form that matches the pattern.
- Tell the user what is blocked and why (the message contains evidence and the gate file path).
- Choose a fundamentally different approach to reach the goal.
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 · +5 lines da768c8af7b5
- 1mo ago First seen · 49 lines · 99 tokens per session scan A a97154e68a0e
dejavu is a skill published in the GitHub repository WhiteBite/opencode-dejavu (6 stars, last pushed 1mo ago), licensed MIT. It adds 99 tokens to every session and 1,122 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
investigate
A structured root-cause investigation method for software bugs. It moves from history and reproduction through tested hypotheses, a focused fix, verification, and a report.
bugfix-triage
A method for locating the cause of a bug before changing code, using a minimal reproduction and step-by-step narrowing of the failing input and code path.
ci-self-heal
A diagnostic guide for analysing failed continuous-integration checks. Continuous integration, or CI, automatically tests and builds code changes; this guide classifies failures and suggests fixes without editing the code.
incident-rca
A diagnostic process for producing a structured root-cause analysis of a serious production incident, such as a P0 or P1 outage. It compares logs, measurements, request traces, code changes, and architecture decisions, and gives ranked explanations and actions.
bug-fix
A structured bug-fixing workflow that takes a problem from reproduction and evidence-based cause finding through approval, implementation, testing, and cleanup.
error-collection
A structured record of confirmed engineering mistakes, their root causes, and checks that prevent them from returning. It uses evidence from real application behavior, not just successful builds.