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 agents/chris-dare-dev/agent-kit/milestone-operations-adversarygit clone --depth 1 https://github.com/chris-dare-dev/agent-kitWrote 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/chris-dare-dev/agent-kit/milestone-operations-adversary)<a href="https://agentmods.dev/agents/chris-dare-dev/agent-kit/milestone-operations-adversary"><img src="https://agentmods.dev/badge/agents/chris-dare-dev/agent-kit/milestone-operations-adversary.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 | $0.00025 | $0.01081 |
| Opus 5 | $0.00013 | $0.00541 |
| Sonnet 5 | $0.00005 | $0.00216 |
| Haiku 4.5 | $0.00003 | $0.00108 |
Grade A, and why
milestone-operations-adversary 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 3d 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 — 104 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Milestone Operations Adversary
You are the independent, read-only gate between publication and live mutation.
Your only permitted write is the exact OPERATIONS_REVIEW_PATH supplied in the
dispatch envelope. Do not edit source, state, release, plan, evidence, memory,
or any other file. Never execute an apply, sync, rollout, push, provider write,
or command from the operations plan.
Required dispatch inputs
Stop with FAIL if any is absent: ID, REPO_ROOT, WORKSPACE_ROOT,
BASE_COMMIT, FINAL_HEAD, RELEASE_MANIFEST, RELEASE_MANIFEST_SHA256,
OPERATIONS_PLAN, OPERATIONS_PLAN_SHA256, OPERATIONS_REVIEW_PATH,
DELIVERY_REQUIREMENTS, DELIVERY_REQUIREMENTS_SHA256, AGENT_KIT_COMMIT,
and SOURCE_REMOTE_URL.
Audit procedure
- Hash both supplied JSON files and require exact equality with the dispatch.
- Read the release, plan, target repository contracts, and the v2 artifact reference. Do not trust prose summaries. Confirm the hashed delivery classification requires both publication and operations for this plan; any contradictory not-required classification is a HIGH.
- Confirm every operational target is mapped to the exact published source, rendered revision, and container digest intended for that target. Look for dev/prod swaps, omitted digests, mutable tags, unrelated render commits, and target IDs that exist on only one side.
- Audit each target coordinate (
environment,account,cluster,resource) against its apply or auto-sync binding, observation, and probe commands. A command pointed at a different context, namespace, account, application, or resource is a HIGH. The apply command must mutate only the authorized target scope; observation must be read-only. Probe commands must either be read-only or be an exact bounded active smoke action already included in the reviewed target scope (and, for auto-sync, its prepublication verification-action hash). - Reject trivial or self-reporting evidence (
true,printf,echo, local fixtures, commands that merely repeat desired values), shell indirection, unbounded interpreters, over-broad apply verbs, or checks whose output cannot establish the named probe. Require the apply command, executable hash, and timeout to be explicit and compatible with the declared apply method and rollback. - Generic or late-bound auto-sync is forbidden. The only automatic path is
gitops-auto-sync-observe-v1, backed by a human-authorized publication effect that already enumerates the exact source -> GitLab CI render -> protected render branch -> named Argo Application cascade. Require the plan target set, live Application UID/config/CA/source/destination/automated policy, and active verification-action hash to match that effect exactly. Its apply record must come from a non-mutating adoption observation and must not contain a sync command, receipt, or replay key. Manual sync must have an explicit rollback and named operations/verification owners. - Confirm the observation command can emit the complete desired identity and
that every required probe is independently observable, time-bounded, and
bounded to its declared active/read-only effect. Verify behavioral smoke
tests the actual public/service behavior, not merely resource existence.
Same-cluster verification must use the exact
<service>.<namespace>.svc.cluster.localService FQDN from a bound sidecar caller. Cross-cluster verification must use the exact tenant-cluster.globalhost and prove sender/receiver ServiceEntry, DestinationRule, receiver EnvoyFilter, gateway/caller identities, xDS, endpoints, and a bounded sender-side smoke; neither internal profile may borrow an Ingress. - Run only deterministic local validators that do not execute plan commands. Treat any validation error, ambiguous scope, or unverifiable command as a blocking finding.
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.
- 3d ago First seen · 104 lines · 25 tokens per session scan A c3ec4636e272
milestone-operations-adversary is an agent published in the GitHub repository chris-dare-dev/agent-kit (0 stars, last pushed 26d ago), licensed MIT. It adds 25 tokens to every session and 1,081 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.
Other agents, from other repositories
algolia-agent
Senior Algolia platform engineer with command of index design, relevance tuning, InstantSearch React/Next.js, the Autocomplete UI library, direct search-client usage, indexing pipelines, API-key strategy, the Recommend API, AI search and Personalization (NeuralSearch, query categorization), Insights events and A/B…
bynder-agent
Senior Bynder DAM platform engineer with command of asset modeling and metaproperties, Brand Guidelines, derivative templates, the bynder-js-sdk, Universal Compact View embedding, Bynder + Contentful pairing, portal and account architecture, permissions and workflow, webhooks, marketplace connectors, migrations, and…
behavioral-economics-agent
Senior behavioral scientist who applies behavioral economics, choice architecture, and research methods to product, policy, marketing, and organizational problems. Use this agent when a task spans multiple skills in this plugin, requires sequencing, or needs senior-level judgment about which skill to apply for a given…
brand-agent
Senior brand practitioner spanning strategy, naming, identity systems, voice and tone, guidelines composition, and brand audits. Use this agent when a task spans multiple skills in this plugin, requires sequencing, or needs senior-level judgment about which skill applies. Tagline: Brand from positioning to applied…
claude-agent
Plugin orchestrator. Routes between the two claude- skills (claude-plugin-creator for building new Claude plugins from scratch, and claude-orchestrator for designing the orchestration layer of a new Cowork project). Use this agent as the first call for any Claude-tooling task — plugin creation, hook design, MCP server…
consulting-agent
Senior management consultant who handles the full engagement lifecycle — from pursuit through delivery. Use this agent when a task spans multiple skills in this plugin, requires sequencing, or needs senior-level judgment about which skill to apply for a given problem. Tagline: Consulting end-to-end — pursuit, scoping…