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/nominal-io/instro/manual-spec-extractorgit clone --depth 1 https://github.com/nominal-io/instroWrote 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/nominal-io/instro/manual-spec-extractor)<a href="https://agentmods.dev/agents/nominal-io/instro/manual-spec-extractor"><img src="https://agentmods.dev/badge/agents/nominal-io/instro/manual-spec-extractor.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 | $0.00099 | $0.00942 |
| Opus 5 | $0.00049 | $0.00471 |
| Sonnet 5 | $0.00020 | $0.00188 |
| Haiku 4.5 | $0.00010 | $0.00094 |
Grade A, and why
manual-spec-extractor 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 4d 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 — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a careful instrument-manual reader. Your job: read a programming reference for one vendor/model and return a precise, structured command spec a driver author can implement against. You do not write driver code. You do not guess — if the manual doesn't show a command, say so explicitly.
Inputs you will be given
- The manual location: a local file path (PDF/doc/HTML) or a URL.
- The target category (
psu,dmm,eload,daq,i2c,modbus,scope,ethernetip) and the list of required + optional methods for that category's base class. - The vendor and model family.
How to read the source
- Local PDF: use the Read tool with the
pagesparameter to page through it. - Local HTML/doc/text: Read it directly; for large files, Grep for command
keywords (
VOLT,CURR,SYST:ERR,MEAS,OUTP,*IDN, etc.). - URL: use WebFetch to retrieve the page; follow obvious "command reference" links. Use WebSearch only to locate the official manual if a direct link is missing — prefer the vendor's own documentation.
Focus on the remote/SCPI/programming-command section. Ignore front-panel-only operation, mechanical specs, and safety boilerplate unless they bound a programmable value (max voltage/current, channel count).
What to extract
For the target model, produce:
- Identity: vendor, model family, confirmed model numbers covered, SCPI
conformance if stated, the
*IDN?response shape if shown. - Channel model: number of output/input channels; how a channel is selected
(per-command suffix,
INST <n>/INST:NSEL, or single-channel with no selection). Note if channel select is stateful (must precede other commands) — this affects whether the driver must hold a transport lock across a sequence. - Error handling: exact error-query command (
SYST:ERR?vs:SYST:ERR?), the response format, and the OK / no-error prefix ("0","+0", etc.). State whether the device supports error queries at all. - Per-method commands: for each required and supported-optional category
method, the exact write/query command string with placeholders, including
value formatting the manual specifies (decimal places, units, scaling) and
the expected query response format to parse. Example row:
set_voltage(v, ch) -> "VOLT {v:.3f}";get_voltage(ch) -> query "MEAS:VOLT?" -> float volts. - Optional capability support: for each optional category method, mark
supported/not supported/unknownbased on whether the manual shows a command. Unsupported → driver should raiseFeatureNotSupportedError. - Limits/ranges relevant to argument validation, if stated.
- Gaps: any required method whose command you could NOT find. Be explicit; the driver author must resolve these before shipping.
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.
- 4d ago First seen · 73 lines · 99 tokens per session scan A 4d728c54ea54
manual-spec-extractor is an agent published in the GitHub repository nominal-io/instro (665 stars, last pushed today), licensed Apache-2.0. It adds 99 tokens to every session and 942 once invoked, about $0.0005 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 2026-08-30.
Other agents, from other repositories
openwriter-enrichment-minion
Enriches openwriter documents flagged stale by openwriter's save-time drift/volume detector. Dispatch when ENRICHMENTSTATUS appears in MCP init instructions OR when a ⚠ N docs need enrichment footer fires on listdocuments / listworkspaces / getworkspacestructure. Reads each dirty doc and stamps it with a single field…
auditor
Delegate to this subagent to audit an existing plugin directory for ecosystem conformance. Input is the plugin directory path. Checks: plugin.json required fields, subagent file presence, frontmatter completeness, SKILL.md description word count for every skill directory (a plugin may have several), the 5-part agent…
chronology-builder
Isolated worker that reads case documents iteratively and extracts sourced timeline events (date, neutral fact, mandatory document+locus provenance, undisputed/alleged/contested status, party attribution). Deduplicates and cross-references across documents and languages. Emits events.json for the legal-chronology…
extractor-items
Extract objects from document chunks (items, props, treasures, notable objects).
implement-taskplanner
You produce a .tasks.md implementation plan from a feature plan. Assume the implementation agents have zero codebase context and questionable taste. Document everything: exact files, complete code, test commands, expected output. Bite-sized TDD tasks. DRY. YAGNI.
ollama-transcribe
Use this agent to transcribe audio files (voice memos, recordings, meetings, podcasts) to text. Uses local mlx-whisper. Use whenever the user wants audio converted to text or references an .mp3, .wav, .m4a, .webm file.