Files
codex/codex-rs/ext/skills/src/render.rs
T
charlesgong-openaiandGitHub 64bdeed9f7 [codex] Preserve skill descriptions outside model context (#29006)
## Why

Skill descriptions are used in model-visible lists: the default
available-skills catalog that supports implicit selection, and the
on-demand `skills.list` tool response used to discover orchestrator
skills. A single overlong description should not consume a
disproportionate share of either list.

Enforcing the 1024-character limit while loading or migrating skills is
the wrong boundary: it rejects otherwise-valid skills and discards
metadata that non-model consumers and full skill reads may need. Skill
metadata and `SKILL.md` content should remain intact; the cap belongs at
model-visible list rendering boundaries.

## What changed

- Preserve full `description` and `metadata.short-description` values
when loading skills.
- Preserve full external-agent command descriptions during
`source-command-*` migration instead of skipping commands solely because
their descriptions exceed 1024 characters.
- Preserve full normalized orchestrator descriptions in the underlying
skills catalog.
- Cap each description at 1024 Unicode characters when rendering the
default available-skills context in `codex-core-skills` and
`codex-skills-extension`.
- Apply the same cap when serializing descriptions in the model-visible
`skills.list` response.
- Render truncated descriptions as 1021 original characters plus `...`.
- Leave explicit `$skill` injection, `skills.read`, underlying metadata,
and on-disk `SKILL.md` files unchanged and full-fidelity.

## Implicit skill selection

Codex injects a bounded catalog containing each implicitly allowed
skill's name, description, and source locator, together with
instructions to use a skill when the task clearly matches its
description. The model makes that semantic choice; after selecting a
skill, it reads the full `SKILL.md` from its filesystem or provider
resource. Explicit `$skill` mentions remain a separate path that injects
the full skill instructions. For orchestrator skills, `skills.list`
provides bounded discovery metadata before `skills.read` returns the
full selected resource.

## Test plan

- `just test -p codex-core-skills`
- `just test -p codex-skills-extension`
- `just test -p codex-external-agent-migration`

The focused regressions verify that overlong metadata is preserved at
load and migration boundaries while default available-skills rendering
and `skills.list` output produce the 1021-character prefix plus `...`.
2026-06-19 12:47:53 -07:00

106 lines
3.5 KiB
Rust

use std::borrow::Cow;
use codex_utils_string::take_bytes_at_char_boundary;
use crate::catalog::SkillCatalog;
use crate::catalog::SkillCatalogEntry;
use crate::catalog::SkillSourceKind;
use crate::fragments::AvailableSkillsInstructions;
const MAX_AVAILABLE_SKILLS_BYTES: usize = 8_000;
const MAX_MAIN_PROMPT_BYTES: usize = 8_000;
const MAX_CATALOG_SKILL_DESCRIPTION_CHARS: usize = 1_024;
const TRUNCATED_SKILL_DESCRIPTION_SUFFIX: &str = "...";
pub(crate) const MAX_SKILL_NAME_BYTES: usize = 256;
pub(crate) const MAX_SKILL_PATH_BYTES: usize = 1_024;
#[tracing::instrument(
level = "trace",
skip_all,
fields(catalog_entry_count = catalog.entries.len())
)]
pub(crate) fn available_skills_fragment(
catalog: &SkillCatalog,
) -> Option<AvailableSkillsInstructions> {
let mut total_bytes = 0usize;
let mut omitted = 0usize;
let mut skill_lines = Vec::new();
for entry in catalog
.entries
.iter()
.filter(|entry| entry.enabled && entry.prompt_visible)
{
let description = entry
.short_description
.as_deref()
.unwrap_or(entry.description.as_str());
let description = truncate_catalog_skill_description(description);
let line = render_skill_line(entry, description.as_ref());
let next_bytes = total_bytes.saturating_add(line.len());
if next_bytes > MAX_AVAILABLE_SKILLS_BYTES {
omitted = omitted.saturating_add(1);
continue;
}
total_bytes = next_bytes;
skill_lines.push(line);
}
if skill_lines.is_empty() {
return None;
}
if omitted > 0 {
let skill_word = if omitted == 1 { "skill" } else { "skills" };
skill_lines.push(format!(
"- {omitted} additional {skill_word} omitted from this bounded skills list."
));
}
Some(AvailableSkillsInstructions::from_skill_lines(skill_lines))
}
pub(crate) fn truncate_catalog_skill_description(description: &str) -> Cow<'_, str> {
if description
.char_indices()
.nth(MAX_CATALOG_SKILL_DESCRIPTION_CHARS)
.is_none()
{
return Cow::Borrowed(description);
}
let prefix_chars = MAX_CATALOG_SKILL_DESCRIPTION_CHARS
.saturating_sub(TRUNCATED_SKILL_DESCRIPTION_SUFFIX.chars().count());
let prefix_end = description
.char_indices()
.nth(prefix_chars)
.map_or(description.len(), |(index, _)| index);
let mut truncated = description[..prefix_end].to_string();
truncated.push_str(TRUNCATED_SKILL_DESCRIPTION_SUFFIX);
Cow::Owned(truncated)
}
fn render_skill_line(entry: &SkillCatalogEntry, description: &str) -> String {
let locator_kind = match &entry.authority.kind {
SkillSourceKind::Host => "file",
SkillSourceKind::Executor => "environment resource",
SkillSourceKind::Orchestrator => "orchestrator resource",
SkillSourceKind::Custom(_) => "custom resource",
};
let name = entry.name.as_str();
let path = entry.rendered_path();
if description.is_empty() {
format!("- {name}: ({locator_kind}: {path})")
} else {
format!("- {name}: {description} ({locator_kind}: {path})")
}
}
pub(crate) fn truncate_main_prompt_contents(contents: &str) -> (String, bool) {
truncate_utf8_to_bytes(contents, MAX_MAIN_PROMPT_BYTES)
}
pub(crate) fn truncate_utf8_to_bytes(contents: &str, max_bytes: usize) -> (String, bool) {
let truncated = take_bytes_at_char_boundary(contents, max_bytes);
(truncated.to_string(), truncated.len() < contents.len())
}