Files
magnus919_agent-skills/dsm5/README.md
Magnus HedemarkGitHubfactory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
c312b36166 fix(skill): split oversized dsm5 references into index + parts (#278)
Agent harnesses truncate file reads around ~60k characters, so the
largest dsm5 reference files (up to 132k chars) were being cut off
mid-file (reported: "The neurodevelopmental file was truncated").

- Split 15 reference files over 50k chars into a small index (original
  filename preserved, so all existing links keep resolving) plus part
  files of <= ~40k chars each, organized by disorder group
- Updated SKILL.md routing rows to point at indexes and read the part
  for the condition; added large-file handling guidance
- Updated dsm5/README.md What You Get table; documented the size
  convention in 00-overview-and-method.md (Maintaining this library)
- Verified: no reference file exceeds 50k chars (66 files), all 466
  relative links resolve, validators pass, lookup.py lists all parts

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
2026-08-04 22:10:15 -04:00

76 lines
3.8 KiB
Markdown
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# dsm5
An evidence-based companion to the DSM-5-TR for serious conversations about mental
health and neurocognitive conditions.
## Why Install This Skill
Mental health questions are everywhere, and most answers online are vibes, not
criteria. This skill grounds those conversations in the real diagnostic framework of
the DSM-5-TR (text revision) so that "could this be X?" becomes a structured
comparison — symptoms weighed against actual criteria, with met, unmet, and unknown
kept separate — instead of a guess. It was built from a copy of the DSM-5-TR text
revision and organized as a searchable reference library, with safety handling and
calibrated language built into every answer.
Who it is for: clinicians and practitioners double-checking criteria, specifiers, or
differentials; patients and family members who want an evidence-based understanding
of a condition and good questions to take to a provider; and anyone who needs to
explain a mental health topic accurately and without stigma.
What it is not: it is not a diagnostic tool, it does not render "you have X"
verdicts, it gives no treatment or medication advice, and it is not the manual
itself. It is a paraphrased companion for orientation and education, and it always
says so.
## What You Get
| Item | What it provides |
|---|---|
| `SKILL.md` | The conversation workflow: safety triage first, question/audience clarification, reference routing, criteria comparison, differential reasoning, and calibrated communication. |
| `references/` (66 files) | Chapter indexes plus part files: foundation (0002), per-chapter references for all 22 DSM-5-TR diagnostic classes (1031), Part III (3233), and cross-cutting differentials (40). Large chapters are split into an index plus focused part files so any agent can read a file in one call. |
| `scripts/lookup.py` | A keyword search across every reference file — finds where a topic lives and which files to read first. |
| `evals/evals.json` | Output-quality test cases that keep the skill's answers honest across safety, criteria fidelity, and plain-language scenarios. |
## Quick Start
No setup, no API keys, no dependencies. Two ways to use it:
- **Browse by chapter.** Open `references/` and pick the file for the condition in
question (for example, `13-depressive-disorders.md` for depression).
- **Search by keyword.** Find where any topic lives:
```sh
python3 scripts/lookup.py 'insomnia'
```
The script prints matching lines grouped by file and a short list of the best
reference files to read. Add `--list` to see every reference file with its
one-line description, or `--json` for machine-readable output.
## Triggers
Load this skill when someone asks about:
- What a set of symptoms "could be" or whether a description matches a known
condition.
- Diagnostic criteria, specifiers, or codes for a mental health condition.
- The difference between two conditions (for example, ADHD vs. anxiety, or delirium
vs. dementia).
- How common a condition is, when it starts, or how it typically runs its course.
- A plain-language explanation of a diagnosis for a patient or family member.
- What questions to bring to a clinician or therapist.
## Requirements
- Python 3 (standard library only) if you want to use `scripts/lookup.py`.
- The reference files, which ship with the skill — no downloads needed.
- For exact verbatim criteria, codes, and recording procedures, consult the official
DSM-5-TR (the user's copy). This skill is a companion, not the authoritative text.
DSM-5-TR is © American Psychiatric Association (2022). This skill is an independent
companion reference; it paraphrases and summarizes the manual for orientation and
education and is not a substitute for the manual, for clinical training, or for
professional evaluation. Always verify criteria and codes against the official
DSM-5-TR before formal use.