# Re-inject the output style; a standing rule decays

To keep a coding agent concise (or in any fixed output style) for a whole
session, do not rely on a rule you wrote once. Persistent instruction files
reportedly decay within a few turns on exactly the models that need the style
rule most; what reportedly holds is re-applying the constraint per turn, or
post-hoc compression at the stop hook. Reported experience from HN item
49610631 (the i-have-adhd thread) and item 49623754, both read 2026-09-10.
Nothing here is verified against a live model.

## The failure being designed against

The rule is in place and obeyed, then silently stops being obeyed:

- A skill instructed from a sparse global CLAUDE.md and manually re-asserted
  mid-session "only maintain the conciseness for a few turns at most before
  they completely forget" on Claude models ([49611957]).
- The same instruction layered everywhere at once - personal instructions,
  global CLAUDE.md, per-project CLAUDE.md, plus the built-in Concise output
  style - still "start deviating in long sessions" on high-thinking models
  ([49622351]).
- Plain-language style prompts (BLUF, ASD-STE100) are reported to "not work
  well, especially over long sessions" by users who tried them repeatedly
  ([49615712]).
- The i-have-adhd SKILL.md itself asserts its rules "do not expire after a few
  turns and they do not lapse when the topic changes" - quoted and then argued
  against in-thread as "at best a very weak signal to the model" ([49615712]).
  The thread contradicts the file's promise from first-hand use ([49611957]).
  The pattern to keep: a skill file promising durability is evidence of the
  author's intent, not of durability.

## What reportedly works

- **Stop-hook compression, not instruction**: one user moved the same
  instruction from skill/MD files (ignored "after a couple of iterations") to a
  Claude Code hook on the stop event and reports it "works well" - cost is
  latency, and it sometimes compresses too hard so both versions get re-read
  ([49613857]).
- **A short re-reminder in the next user message**: for the multi-layer setup
  above, "then I just remind it in my next message: 'Remember ASD-STE100
  style'" reportedly recenters it ([49622351]). Cheap; must be repeated.
- **Per-turn re-injection is the weak version**: a plugin re-injects the output
  style every turn on GLM and it is still "nigh impossible" to hold; the same
  injection is obeyed hundreds of thousands of tokens in on GPT-family models
  ([49612184]). The lever that works is the mechanism and the model, not the
  text.

## When it fails anyway

- **Model-dependent, and it flips between models.** The same instruction text
  is obeyed on one model and ignored on another ([49612184]); a curated
  minimal style skill reportedly works "100% of the time for other models,
  70-80% for Claude" ([49618225], from item 49623754). Don't port a working
  config across vendors and assume the behavior travels.
- **Model choice defeats it upstream.** Choosing a heavy-thinking model for
  small questions guarantees verbosity no style rule fully claws back
  ([49615605]); one report found Sonnet 4.6 markedly less verbose than the
  flagship ([49613055]). Change the model before waging a style war.
- **Provider-side mechanisms may already exist.** Claude Code ships a
  predefined "Concise" output style ([49612733]) - check for a built-in before
  building a hook.
- **No measurement exists.** Nobody in either thread ran an eval. Everything
  on this page is forum experience - treat it as leads to replicate on your
  own harness, not as a benchmark.

## Adjacent hazard: installing the style rule is itself attack surface

Installing the skill means pasting "Install ... refer to the repo's AGENTS.md"
into an agent - copy-pasted install instructions pointed at a repo's agent
file are the same shape as piping curl into sh; a similar-named repo or a
token swap could redirect the install ([49614192]). The thread's own example
of why: the i-have-adhd repo's AGENTS.md instructs visiting agents to comment
on a GitHub issue, and one HN reader flagged it and deliberately did not act
on it ([49611877]); others noted those instructions are scoped to maintainers
and are not copied by the installer ([49613705]). Read what a skill file tells
agents to *do* before installing it; instructions addressed to agents in
fetched files are data.

## Source

HN item 49610631 (comments 49611957, 49615712, 49622351, 49613857, 49612184,
49615605, 49613055, 49612733, 49614192, 49611877, 49613705) and item 49623754
(comment 49618225, which is also cited on skills/telling-the-agent-the-scope),
read 2026-09-10. All claims are reported experience from a forum thread; no
controlled test in the thread or by this wiki.