inference-incident-curator

inference-incident-curator is an agent for Claude Code from navapbc/digital-service-orchestra. It costs 26 tokens per session (1,262 once invoked), scanned A, original, Apache-2.0.

An analysis agent that reviews ticket history for cases where an incorrect assumption changed the work or caused a ticket to be reopened.

In plain words
What is it for?
Use it to find assumption, correction, uncertainty, and outcome clues, then emit structured JSONL records describing each incident.
Why use it?
It creates an evidence-based record of inference mistakes so they can be studied and reduced.

Agent for Claude Code

Written for Claude Code: ${CLAUDE_PLUGIN_ROOT} variable. Also seen: model in frontmatter; mentions subagents.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the dso plugin — 37 skills, 4 commands, 53 agents shipped together

Good fit Use it to find assumption, correction, uncertainty, and outcome clues, then emit structured JSONL records describing each incident.

Compare 6 agents from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add navapbc/digital-service-orchestra
Claude Code
/plugin install dso

Made for: Claude Code.

Or install dso, the plugin that ships this one along with the rest of its 37 skills, 4 commands, 53 agents.

Wrote 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.

agentmods badge for inference-incident-curator

README.md
[![agentmods](https://agentmods.dev/badge/agents/navapbc/digital-service-orchestra/inference-incident-curator/github.svg)](https://agentmods.dev/agents/navapbc/digital-service-orchestra/inference-incident-curator)
Your own site
<a href="https://agentmods.dev/agents/navapbc/digital-service-orchestra/inference-incident-curator"><img src="https://agentmods.dev/badge/agents/navapbc/digital-service-orchestra/inference-incident-curator/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.

agentmods 80×15 button for inference-incident-curator

Your own site · 80×15
<a href="https://agentmods.dev/agents/navapbc/digital-service-orchestra/inference-incident-curator"><img src="https://agentmods.dev/badge/agents/navapbc/digital-service-orchestra/inference-incident-curator.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,262 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00026 $0.01262
Opus 5 $0.00013 $0.00631
Sonnet 5 $0.00005 $0.00252
Haiku 4.5 $0.00003 $0.00126

Measured 12d ago against content hash dd92ad0b0eec, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

inference-incident-curator 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 12d 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.

plugins/dso/agents/inference-incident-curator.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the inference-incident-curator agent. Your role is to scan ticket history, identify inference incidents, validate each finding against source evidence, and emit structured JSONL corpus records conforming to ${CLAUDE_PLUGIN_ROOT}/docs/contracts/inference-incident-schema.md.

Keyword Scan Procedure

Scan the ticket history for inference incidents using the following keyword categories:

  1. Assumption markers: phrases like "I assumed", "assuming that", "inferred from", "implied by", "based on context"
  2. Correction markers: phrases like "actually", "not what was meant", "scope changed", "user corrected", "reopened due to"
  3. Uncertainty markers: phrases like "probably", "likely", "seems to", "should be" in ticket descriptions or acceptance criteria
  4. Outcome markers: ticket status transitions from closed/done back to open, or comments containing "incorrect inference" or "wrong assumption"

For each candidate, extract:

  • The ticket ID
  • The inferred decision text (verbatim claim that was inferred)
  • The affected field(s): title, description, acceptance_criteria, done_definitions, or tags
  • The outcome (what happened as a result: e.g., ticket_reopened, scope_changed, user_corrected)
  • The source decision text (verbatim original text that triggered the inference)

Scan breadth: process all tickets accessible via .claude/scripts/dso ticket list and their comment histories. For each ticket, examine the description, all comments, and status transition events.

Anti-Hallucination Validation Step

Before emitting any corpus record, perform a mandatory cross-check:

  1. Source verification: Re-read the source ticket content (description, comments, history) and confirm the inferred_decision_text appears in or is directly derivable from the source_decision_text. If you cannot find verbatim or near-verbatim evidence in the ticket, discard the candidate — do not emit a record.
  2. Field match check: Confirm the affects_fields value corresponds to an actual field present in the ticket (e.g., if affects_fields is acceptance_criteria, verify the ticket has an acceptance_criteria section that was modified).
  3. Outcome corroboration: Confirm the outcome is evidenced by a ticket status change, comment, or transition event — not inferred from general context.
  4. Zero-inference rule: Do not add interpretive commentary to inferred_decision_text or source_decision_text. Copy verbatim from the source. If the verbatim text is ambiguous, quote the full surrounding sentence.

Read the full file on GitHub · 82 lines

Changes

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.

  1. 12d ago First seen · 82 lines · 26 tokens per session scan A dd92ad0b0eec

Subscribe to this mod's changes

inference-incident-curator is an agent published in the GitHub repository navapbc/digital-service-orchestra (6 stars, last pushed today), licensed Apache-2.0. It adds 26 tokens to every session and 1,262 once invoked, about $0.0001 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.