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 smixs/iva-agent --skill self-mapgit clone --depth 1 https://github.com/smixs/iva-agentWrote 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/smixs/iva-agent/self-map)<a href="https://agentmods.dev/skills/smixs/iva-agent/self-map"><img src="https://agentmods.dev/badge/skills/smixs/iva-agent/self-map/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/smixs/iva-agent/self-map"><img src="https://agentmods.dev/badge/skills/smixs/iva-agent/self-map.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 | $0.00085 | $0.03800 |
| Opus 5.5 | $0.00034 | $0.01520 |
| Sonnet 5.5 | $0.00017 | $0.00760 |
| Haiku 4.5 | $0.00009 | $0.00380 |
Grade A, and why
self-map 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 yesterday.
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 — 260 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Self-map
Paths are relative to the bash folder — the running Version
(~/iva/versions/<v>; ~/iva/current points at it). Your own code is here, not
in ~/iva. Data is $ASSISTANT_DATA_DIR (default ./data). Everything below
only reads.
Keep the output small
bash gives you only the last 30 000 characters of each stream; read_file
stops at 24 000. Never cat a day file of the Trace or data/usage.jsonl:
they run to megabytes. A long turn — print only what is around its failures:
iva trace show <session>/<turn> | grep -B4 -A2 -e 'failure=' -e 'turn.failed' -e 'step.failed'
Never in a turn without the owner
iva doctor— blocked: its repairs restart iva.service.iva diagnose— read-only; in a turn only throughreport-problem.iva logs,iva trace tail— they follow forever and end only on the bash timeout. Usejournalctl … --no-pager -n Nandiva trace show.iva proactive on|off|set,iva jobs ack— they change state; only on the owner's word.
One turn = session + turn
turn_N starts again at turn_0 in every session, so a turn is named by both:
<session>/<turn>, for example wrun_01M43YEKCW0FRANBS1E510CWX5/turn_0. Nobody
tells you your own session; find the turn from what you have:
- A line of the failure list in your prompt — it ends with the command; run it.
- A recent turn —
iva trace showlists the last 20 with their<session>/<turn>;iva trace show lastis the last turn of the model. - A chat message —
iva trace show tg:<chat>:<message>. - A time or words —
tsis UTC, the day file is the local day:grep -l '<words>' data/trace/*.jsonl, thengrep -h '<words>' data/trace/<day>.jsonl | cut -c1-200showsturnandsession.
A turn without a chat (Watch, Brief, Insight, a Reminder) starts with an
eve.message.received whose prompt begins Watch:, Brief:, Insight: or
Reminder. The lines that sent its answer (gate.outbound, outbox.*) carry
no session: they are the sends right after its turn.completed.
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.
- yesterday Changed · +8 lines 3845615fef55
- 4d ago Changed · +64 lines · +3 tokens per session 7705bffedb4b
- 7d ago First seen · 188 lines · 82 tokens per session scan A c8b7dd331e83
self-map is a skill published in the GitHub repository smixs/iva-agent (240 stars, last pushed today), licensed MIT. It adds 85 tokens to every session and 3,800 once invoked, about $0.0003 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 2026-10-09.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior. 4-phase root cause investigation — NO fixes without understanding the problem first.
code-review
Performs structured code review on a file or directory.
regex
Write, test, and debug regular expressions for pattern matching, search, and text extraction across programming languages. Use when the user asks to match patterns in text, validate input formats (email, URL, IP, phone, date), extract data with capture groups, build search patterns with lookahead or lookbehind, or…
code-analysis
LSP-based code analysis via mcpls MCP server. Use for compiler-accurate type inspection, go-to-definition, find-all-references, diagnostics, call hierarchy, workspace symbol search, code actions, rename refactoring, and document formatting. Provides real compiler errors — not guesses. Keywords: LSP, language server…
network
Network diagnostics, DNS lookup, port scanning, and connectivity testing. Use when the user asks to check if a host is reachable, trace network routes, resolve DNS records, inspect open ports, debug HTTP requests, check SSL/TLS certificates, test bandwidth, diagnose network latency, or troubleshoot connectivity issues…
self-ops
Runtime self-diagnosis knowledge: database layout, observability APIs, timeout hierarchy, restart procedure, failure replay. Use when the user asks to debug this runtime itself — inspect its database, traces, prompt snapshots, timeouts, service health, or configuration.