harn_skills/
lib.rs

1//! Embedded Harn skill corpus.
2//!
3//! This crate exposes the bundled corpus as metadata plus `SKILL.md`
4//! bodies. CLI commands that enumerate, dump, or install these skills
5//! are layered above this foundation.
6
7use std::collections::BTreeMap;
8use std::env;
9use std::fmt;
10use std::fs;
11use std::io;
12use std::path::{Path, PathBuf};
13use std::sync::OnceLock;
14
15/// Environment override for canonical Harn skill discovery.
16pub const HARN_SKILLS_DIR_ENV: &str = "HARN_SKILLS_DIR";
17
18/// Frontmatter fields embedded with each bundled skill.
19#[derive(Debug, Clone, Copy, PartialEq, Eq)]
20pub struct SkillFrontmatter {
21    pub name: &'static str,
22    pub short: &'static str,
23    pub description: &'static str,
24    pub when_to_use: Option<&'static str>,
25}
26
27/// A single skill embedded into the Harn build.
28#[derive(Debug, Clone, Copy, PartialEq, Eq)]
29pub struct EmbeddedSkill {
30    pub name: &'static str,
31    pub frontmatter: SkillFrontmatter,
32    pub body: &'static str,
33    /// The full original SKILL.md source — frontmatter delimiter,
34    /// frontmatter block, blank line, and body — exactly as embedded.
35    /// Use this when round-tripping a skill back to disk so the dumped
36    /// copy is byte-identical to the binary's canonical record.
37    pub source: &'static str,
38}
39
40/// Owned frontmatter fields loaded from a `SKILL.md` on disk.
41#[derive(Debug, Clone, PartialEq, Eq)]
42pub struct DiskSkillFrontmatter {
43    pub name: String,
44    pub short: String,
45    pub description: String,
46    pub when_to_use: Option<String>,
47}
48
49/// A single skill discovered recursively from `HARN_SKILLS_DIR`.
50#[derive(Debug, Clone, PartialEq, Eq)]
51pub struct DiskSkill {
52    pub name: String,
53    pub frontmatter: DiskSkillFrontmatter,
54    pub body: String,
55    pub source: String,
56    pub path: PathBuf,
57}
58
59/// The active canonical corpus used by `harn skills list/get`.
60#[derive(Debug, Clone, PartialEq, Eq)]
61pub enum SkillCorpus {
62    Embedded(&'static [EmbeddedSkill]),
63    Disk(Vec<DiskSkill>),
64}
65
66impl SkillCorpus {
67    pub fn is_disk(&self) -> bool {
68        matches!(self, Self::Disk(_))
69    }
70
71    pub fn len(&self) -> usize {
72        match self {
73            Self::Embedded(skills) => skills.len(),
74            Self::Disk(skills) => skills.len(),
75        }
76    }
77
78    pub fn is_empty(&self) -> bool {
79        self.len() == 0
80    }
81}
82
83/// Error returned when disk skill discovery finds malformed files.
84#[derive(Debug)]
85pub enum SkillDiscoveryError {
86    Io {
87        path: PathBuf,
88        source: io::Error,
89    },
90    MissingFrontmatter {
91        path: PathBuf,
92    },
93    MissingField {
94        path: PathBuf,
95        field: &'static str,
96    },
97    DuplicateName {
98        name: String,
99        first: PathBuf,
100        second: PathBuf,
101    },
102}
103
104impl fmt::Display for SkillDiscoveryError {
105    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
106        match self {
107            Self::Io { path, source } => write!(f, "{}: {source}", path.display()),
108            Self::MissingFrontmatter { path } => {
109                write!(f, "{}: missing SKILL.md frontmatter", path.display())
110            }
111            Self::MissingField { path, field } => {
112                write!(f, "{}: missing `{field}` frontmatter field", path.display())
113            }
114            Self::DuplicateName {
115                name,
116                first,
117                second,
118            } => write!(
119                f,
120                "duplicate skill `{name}` in {} and {}",
121                first.display(),
122                second.display()
123            ),
124        }
125    }
126}
127
128impl std::error::Error for SkillDiscoveryError {}
129
130const SOURCES: &[&str] = &[
131    include_str!("corpus/harn-agent/SKILL.md"),
132    include_str!("corpus/harn-diagnostics/SKILL.md"),
133    include_str!("corpus/harn-language/SKILL.md"),
134    include_str!("corpus/harn-orchestration/SKILL.md"),
135    include_str!("corpus/harn-providers/SKILL.md"),
136    include_str!("corpus/harn-testing/SKILL.md"),
137    include_str!("corpus/harn-tracing/SKILL.md"),
138];
139
140static EMBEDDED_SKILLS: OnceLock<Box<[EmbeddedSkill]>> = OnceLock::new();
141
142/// Return every skill bundled into this build.
143pub fn list_embedded_skills() -> &'static [EmbeddedSkill] {
144    EMBEDDED_SKILLS
145        .get_or_init(|| SOURCES.iter().map(|source| parse_skill(source)).collect())
146        .as_ref()
147}
148
149/// Return one bundled skill by canonical skill name.
150pub fn get_embedded_skill(name: &str) -> Option<&'static EmbeddedSkill> {
151    list_embedded_skills()
152        .iter()
153        .find(|skill| skill.name == name)
154}
155
156/// Return the active canonical corpus. `HARN_SKILLS_DIR` wins only
157/// when it contains at least one recursively discovered `SKILL.md`;
158/// otherwise callers fall back to the embedded corpus.
159pub fn resolve_skill_corpus_from_env() -> Result<SkillCorpus, SkillDiscoveryError> {
160    let Ok(dir) = env::var(HARN_SKILLS_DIR_ENV) else {
161        return Ok(SkillCorpus::Embedded(list_embedded_skills()));
162    };
163    if dir.trim().is_empty() {
164        return Ok(SkillCorpus::Embedded(list_embedded_skills()));
165    }
166
167    let skills = list_disk_skills(dir)?;
168    if skills.is_empty() {
169        Ok(SkillCorpus::Embedded(list_embedded_skills()))
170    } else {
171        Ok(SkillCorpus::Disk(skills))
172    }
173}
174
175/// Recursively discover `SKILL.md` files under `root`.
176///
177/// A missing root is treated as an empty disk corpus so
178/// `HARN_SKILLS_DIR` can fall back cleanly to embedded skills.
179pub fn list_disk_skills(root: impl AsRef<Path>) -> Result<Vec<DiskSkill>, SkillDiscoveryError> {
180    let root = root.as_ref();
181    if !root.exists() {
182        return Ok(Vec::new());
183    }
184
185    let mut paths = Vec::new();
186    collect_skill_paths(root, &mut paths)?;
187    paths.sort();
188
189    let mut by_name: BTreeMap<String, DiskSkill> = BTreeMap::new();
190    for path in paths {
191        let skill = parse_disk_skill(&path)?;
192        if let Some(first) = by_name.get(&skill.name) {
193            return Err(SkillDiscoveryError::DuplicateName {
194                name: skill.name,
195                first: first.path.clone(),
196                second: path,
197            });
198        }
199        by_name.insert(skill.name.clone(), skill);
200    }
201
202    Ok(by_name.into_values().collect())
203}
204
205fn parse_skill(source: &'static str) -> EmbeddedSkill {
206    let (frontmatter, body) = split_frontmatter(source);
207    let frontmatter = parse_frontmatter(frontmatter);
208    EmbeddedSkill {
209        name: frontmatter.name,
210        frontmatter,
211        body,
212        source,
213    }
214}
215
216fn collect_skill_paths(dir: &Path, out: &mut Vec<PathBuf>) -> Result<(), SkillDiscoveryError> {
217    let entries = fs::read_dir(dir).map_err(|source| SkillDiscoveryError::Io {
218        path: dir.to_path_buf(),
219        source,
220    })?;
221    for entry in entries {
222        let entry = entry.map_err(|source| SkillDiscoveryError::Io {
223            path: dir.to_path_buf(),
224            source,
225        })?;
226        let path = entry.path();
227        let file_type = entry
228            .file_type()
229            .map_err(|source| SkillDiscoveryError::Io {
230                path: path.clone(),
231                source,
232            })?;
233        if file_type.is_dir() {
234            collect_skill_paths(&path, out)?;
235        } else if file_type.is_file() && entry.file_name() == "SKILL.md" {
236            out.push(path);
237        }
238    }
239    Ok(())
240}
241
242fn parse_disk_skill(path: &Path) -> Result<DiskSkill, SkillDiscoveryError> {
243    let source = fs::read_to_string(path).map_err(|source| SkillDiscoveryError::Io {
244        path: path.to_path_buf(),
245        source,
246    })?;
247    let (frontmatter, body) =
248        split_disk_frontmatter(&source).ok_or_else(|| SkillDiscoveryError::MissingFrontmatter {
249            path: path.to_path_buf(),
250        })?;
251    let frontmatter = parse_disk_frontmatter(path, frontmatter)?;
252    Ok(DiskSkill {
253        name: frontmatter.name.clone(),
254        frontmatter,
255        body: body.to_string(),
256        source,
257        path: path.to_path_buf(),
258    })
259}
260
261fn split_disk_frontmatter(source: &str) -> Option<(&str, &str)> {
262    split_frontmatter_parts(source)
263}
264
265fn parse_disk_frontmatter(
266    path: &Path,
267    frontmatter: &str,
268) -> Result<DiskSkillFrontmatter, SkillDiscoveryError> {
269    let mut name = None;
270    let mut short = None;
271    let mut description = None;
272    let mut when_to_use = None;
273
274    for line in frontmatter.lines() {
275        let Some((key, value)) = line.split_once(':') else {
276            continue;
277        };
278        let value = value.trim().to_string();
279        match key {
280            "name" => name = Some(value),
281            "short" => short = Some(value),
282            "description" => description = Some(value),
283            "when_to_use" => when_to_use = Some(value),
284            _ => {}
285        }
286    }
287
288    Ok(DiskSkillFrontmatter {
289        name: require_disk_field(path, name, "name")?,
290        short: short.unwrap_or_default(),
291        description: require_disk_field(path, description, "description")?,
292        when_to_use,
293    })
294}
295
296fn require_disk_field(
297    path: &Path,
298    value: Option<String>,
299    field: &'static str,
300) -> Result<String, SkillDiscoveryError> {
301    value.ok_or_else(|| SkillDiscoveryError::MissingField {
302        path: path.to_path_buf(),
303        field,
304    })
305}
306
307fn split_frontmatter(source: &'static str) -> (&'static str, &'static str) {
308    let Some((after_open, line_ending)) = split_opening_frontmatter(source) else {
309        panic!("embedded skill source is missing opening frontmatter delimiter");
310    };
311    let Some((frontmatter, body)) = split_closing_frontmatter(after_open, line_ending) else {
312        panic!("embedded skill source is missing closing frontmatter delimiter");
313    };
314    (frontmatter, body)
315}
316
317fn split_frontmatter_parts(source: &str) -> Option<(&str, &str)> {
318    let (after_open, line_ending) = split_opening_frontmatter(source)?;
319    split_closing_frontmatter(after_open, line_ending)
320}
321
322fn split_opening_frontmatter(source: &str) -> Option<(&str, &str)> {
323    if let Some(after_open) = source.strip_prefix("---\n") {
324        Some((after_open, "\n"))
325    } else if let Some(after_open) = source.strip_prefix("---\r\n") {
326        Some((after_open, "\r\n"))
327    } else {
328        None
329    }
330}
331
332fn split_closing_frontmatter<'a>(
333    after_open: &'a str,
334    line_ending: &str,
335) -> Option<(&'a str, &'a str)> {
336    let close = format!("{line_ending}---{line_ending}");
337    let close_offset = after_open.find(&close)?;
338    Some((
339        &after_open[..close_offset],
340        &after_open[close_offset + close.len()..],
341    ))
342}
343
344fn parse_frontmatter(frontmatter: &'static str) -> SkillFrontmatter {
345    let mut name = None;
346    let mut short = None;
347    let mut description = None;
348    let mut when_to_use = None;
349
350    for line in frontmatter.lines() {
351        let Some((key, value)) = line.split_once(':') else {
352            continue;
353        };
354        let value = value.trim();
355        match key {
356            "name" => name = Some(value),
357            "short" => short = Some(value),
358            "description" => description = Some(value),
359            "when_to_use" => when_to_use = Some(value),
360            _ => {}
361        }
362    }
363
364    SkillFrontmatter {
365        name: name.expect("embedded skill frontmatter is missing `name`"),
366        short: short.expect("embedded skill frontmatter is missing `short`"),
367        description: description.expect("embedded skill frontmatter is missing `description`"),
368        when_to_use,
369    }
370}
371
372#[cfg(test)]
373mod tests {
374    use super::*;
375    use std::collections::BTreeSet;
376    use tempfile::TempDir;
377
378    #[test]
379    fn lists_expected_initial_corpus() {
380        let skills = list_embedded_skills();
381        let names: Vec<&str> = skills.iter().map(|skill| skill.name).collect();
382        assert_eq!(
383            names,
384            [
385                "harn-agent",
386                "harn-diagnostics",
387                "harn-language",
388                "harn-orchestration",
389                "harn-providers",
390                "harn-testing",
391                "harn-tracing",
392            ]
393        );
394        assert_eq!(skills.len(), SOURCES.len());
395    }
396
397    #[test]
398    fn can_fetch_harn_language_skill() {
399        let skill = get_embedded_skill("harn-language").expect("harn-language skill is embedded");
400        assert_eq!(skill.frontmatter.name, "harn-language");
401        assert!(skill.body.contains("Harn language"));
402    }
403
404    #[test]
405    fn skills_have_unique_names_and_body_only_content() {
406        let mut names = BTreeSet::new();
407        for skill in list_embedded_skills() {
408            assert_eq!(skill.name, skill.frontmatter.name);
409            assert!(names.insert(skill.name), "duplicate skill {}", skill.name);
410            assert!(
411                !skill.body.trim().is_empty(),
412                "{} body is empty",
413                skill.name
414            );
415            assert!(
416                !skill.body.trim_start().starts_with("---"),
417                "{} body includes frontmatter",
418                skill.name
419            );
420        }
421    }
422
423    #[test]
424    fn skills_are_sorted_by_name() {
425        let names: Vec<&str> = list_embedded_skills()
426            .iter()
427            .map(|skill| skill.name)
428            .collect();
429        let mut sorted = names.clone();
430        sorted.sort_unstable();
431        assert_eq!(names, sorted);
432    }
433
434    #[test]
435    fn source_round_trips_to_frontmatter_and_body() {
436        for skill in list_embedded_skills() {
437            assert!(
438                split_frontmatter_parts(skill.source).is_some(),
439                "{} source missing opening fence",
440                skill.name
441            );
442            assert!(
443                skill.source.ends_with(skill.body),
444                "{} source must end with the body so dump output is byte-stable",
445                skill.name
446            );
447            assert!(
448                skill.source.contains(&format!("name: {}\n", skill.name)),
449                "{} source missing canonical name field",
450                skill.name
451            );
452        }
453    }
454
455    #[test]
456    fn frontmatter_split_accepts_crlf_sources() {
457        let source = "---\r\nname: crlf\r\n---\r\n# Body\r\n";
458        let (frontmatter, body) = split_frontmatter_parts(source).expect("CRLF frontmatter");
459        assert_eq!(frontmatter, "name: crlf");
460        assert_eq!(body, "# Body\r\n");
461    }
462
463    #[test]
464    fn frontmatter_split_rejects_missing_closing_fence() {
465        assert!(split_frontmatter_parts("---\nname: missing\n# Body\n").is_none());
466    }
467
468    #[test]
469    fn embedded_corpus_stays_within_binary_budget() {
470        let bytes: usize = SOURCES.iter().map(|source| source.len()).sum();
471        assert!(
472            bytes <= 200 * 1024,
473            "embedded corpus is {bytes} bytes, expected <= 200 KiB"
474        );
475    }
476
477    #[test]
478    fn skill_bodies_are_focused_and_not_placeholders() {
479        let expectations = [
480            ("harn-agent", ["agent_loop", "session id", "approval"]),
481            ("harn-diagnostics", ["diagnostic", "repair", "conformance"]),
482            ("harn-language", ["quickref", "type", "conformance"]),
483            ("harn-orchestration", ["agent_loop", "workflow", "host"]),
484            ("harn-providers", ["llm_call", "provider", "schema"]),
485            (
486                "harn-testing",
487                ["conformance", "deterministic", "mock_time"],
488            ),
489            ("harn-tracing", ["replay", "receipts", "transcript"]),
490        ];
491
492        for (name, terms) in expectations {
493            let skill = get_embedded_skill(name).expect("expected embedded skill");
494            let body = skill.body.to_ascii_lowercase();
495            assert!(
496                !body.contains("embedded stub") && !body.contains("placeholder"),
497                "{name} should contain real guidance, not stub wording"
498            );
499            for term in terms {
500                assert!(
501                    body.contains(term),
502                    "{name} body should mention focused term `{term}`"
503                );
504            }
505        }
506    }
507
508    #[test]
509    fn skill_bodies_match_split_skill_contract() {
510        for skill in list_embedded_skills() {
511            let lines = skill.body.lines().count();
512            assert!(
513                lines >= 80,
514                "{} body is {lines} lines, expected at least 80",
515                skill.name
516            );
517            assert!(
518                lines <= 300,
519                "{} body is {lines} lines, expected at most 300",
520                skill.name
521            );
522        }
523    }
524
525    #[test]
526    fn skill_cross_links_resolve_to_embedded_skills() {
527        let names: BTreeSet<&str> = list_embedded_skills()
528            .iter()
529            .map(|skill| skill.name)
530            .collect();
531        for skill in list_embedded_skills() {
532            for reference in bracketed_skill_references(skill.body) {
533                assert!(
534                    names.contains(reference),
535                    "{} links to unknown embedded skill [[{}]]",
536                    skill.name,
537                    reference
538                );
539            }
540        }
541    }
542
543    #[test]
544    fn diagnostics_skill_mentions_all_code_categories() {
545        let skill = get_embedded_skill("harn-diagnostics").expect("diagnostics skill");
546        for category in [
547            "TYP", "PAR", "NAM", "CAP", "LLM", "ORC", "STD", "PRM", "MOD", "LNT", "FMT", "IMP",
548            "OWN", "RCV", "MAT",
549        ] {
550            assert!(
551                skill.body.contains(&format!("`{category}`")),
552                "harn-diagnostics should mention diagnostic category `{category}`"
553            );
554        }
555    }
556
557    #[test]
558    fn disk_discovery_finds_recursive_skill_files_sorted_by_name() {
559        let temp = TempDir::new().expect("temp dir");
560        write_skill(
561            &temp.path().join("zeta").join("SKILL.md"),
562            "zeta-skill",
563            "Zeta",
564        );
565        write_skill(
566            &temp.path().join("nested").join("alpha").join("SKILL.md"),
567            "alpha-skill",
568            "Alpha",
569        );
570
571        let skills = list_disk_skills(temp.path()).expect("discover disk skills");
572        let names: Vec<&str> = skills.iter().map(|skill| skill.name.as_str()).collect();
573        assert_eq!(names, ["alpha-skill", "zeta-skill"]);
574        assert_eq!(skills[0].frontmatter.description, "Alpha description");
575        assert!(skills[0].body.contains("Alpha body"));
576    }
577
578    #[test]
579    fn disk_discovery_treats_missing_root_as_empty() {
580        let temp = TempDir::new().expect("temp dir");
581        let skills = list_disk_skills(temp.path().join("missing")).expect("discover disk skills");
582        assert!(skills.is_empty());
583    }
584
585    #[test]
586    fn disk_discovery_rejects_duplicate_skill_names() {
587        let temp = TempDir::new().expect("temp dir");
588        write_skill(
589            &temp.path().join("one").join("SKILL.md"),
590            "same-skill",
591            "One",
592        );
593        write_skill(
594            &temp.path().join("two").join("SKILL.md"),
595            "same-skill",
596            "Two",
597        );
598
599        let error = list_disk_skills(temp.path()).expect_err("duplicate name should fail");
600        assert!(
601            error.to_string().contains("duplicate skill `same-skill`"),
602            "unexpected error: {error}"
603        );
604    }
605
606    fn write_skill(path: &Path, name: &str, label: &str) {
607        fs::create_dir_all(path.parent().expect("skill parent")).expect("create skill parent");
608        fs::write(
609            path,
610            format!(
611                "---\nname: {name}\nshort: {label} short\ndescription: {label} description\n---\n# {label}\n\n{label} body\n"
612            ),
613        )
614        .expect("write SKILL.md");
615    }
616
617    fn bracketed_skill_references(body: &str) -> Vec<&str> {
618        let mut references = Vec::new();
619        let mut rest = body;
620        while let Some(start) = rest.find("[[") {
621            rest = &rest[start + 2..];
622            let Some(end) = rest.find("]]") else {
623                break;
624            };
625            references.push(&rest[..end]);
626            rest = &rest[end + 2..];
627        }
628        references
629    }
630}
harn_skills/lib.rs

harn_skills/
lib.rs
