Guard source and build carrying cost

This commit is contained in:
Meta Vens 2026-08-23 12:00:43 +02:00
commit 8c079d9fdd
3 changed files with 79 additions and 4 deletions

View file

@ -82,7 +82,7 @@ machine-spirit awake.
- Handoff summary: `F:\Projects\Epiphany\notes\fresh-workspace-handoff.md`
- Epiphany algorithmic map: `F:\Projects\Epiphany\notes\epiphany-current-algorithmic-map.md`
- Epiphany safety architecture: `F:\Projects\Epiphany\notes\epiphany-safety-architecture.md`
- State CLI: `cargo run --manifest-path F:\Projects\Epiphany\epiphany-core\Cargo.toml --bin epiphany-state -- ...`
- State CLI: `cargo run --manifest-path F:\Projects\Epiphany\Cargo.toml --bin epiphany-state -- ...`
- Pre-compaction helper: `cargo run --manifest-path F:\Projects\Epiphany\epiphany-core\Cargo.toml --bin epiphany-prepare-compaction -- ...`
## Useful Commands
@ -96,7 +96,7 @@ name the shared target explicitly:
```powershell
& 'C:\Users\Meta\.cargo-target-codex\debug\epiphany-state.exe' status
$env:CARGO_TARGET_DIR='C:\Users\Meta\.cargo-target-codex'
cargo run --manifest-path .\epiphany-core\Cargo.toml --bin epiphany-state -- add-evidence --type research --status ok --note '...'
cargo run --manifest-path .\Cargo.toml --bin epiphany-state -- add-evidence --type research --status ok --note '...'
cargo run --manifest-path .\epiphany-core\Cargo.toml --bin epiphany-prepare-compaction --
```
@ -219,10 +219,50 @@ operator-enginseer, not the implementation servitor.
run contaminated for supervision purposes, and continue from sealed
projections instead of letting the worker's stream steer the supervisor.
## Source And Build Economy
- Verified capability is the asset. Epiphany-owned source, tests, schemas,
dependencies, executable targets, and compiler artifacts are carrying costs.
Minimize authority surfaces and build fan-out, not line count; readable code
and consequence-bearing verifier reach remain invariants.
- The target invariant is one Cargo package owner per production entrypoint.
`epiphany-core` owns only the `epiphany-prepare-compaction` maintenance
binary; runtime entrypoints are converging on `epiphany-release-bundle`.
Known violations still under repair are `epiphany-model-runtime`,
`epiphany-persona-service`, and `epiphany-tool-mcp-runtime`, whose sources are
also admitted by their leaf package manifests. Treat those duplicates as cut
work, not precedent. Do not point multiple packages at the same binary source
or preserve a binary because a stale manifest, smoke command, or historical
document still names it.
- Admit a new executable only when a live consumer requires an independent
process, privilege boundary, lifecycle, or deployment contract. Tests,
fixtures, probes, and one-off migrations belong in existing library or
integration-test surfaces unless they must cross such a boundary.
- Local Cargo checks must name one package and a focused library, integration
test, or binary target. Do not run local workspace-wide, `--all-targets`, or
otherwise unspecified multi-binary builds without explicit operator
authorization. Idunn owns release-wide build execution. Before an authorized
broad build, inventory the target graph, output root, current output size,
free disk, expected budget, and terminal retention.
- An interrupted build remains unsettled work. Before retrying, inventory every
output root it may have written, including absolute path, size, last write,
owning run or revision, live process status, and retention decision. Settle
that run before creating another target root; cleanup machinery is not a
substitute for deleting unowned source or target fan-out.
## Verification Guardrails
- Use focused checks for the surface being changed instead of defaulting to a
whole-repo ritual. Not every prayer needs a cathedral.
- For Epiphany control-plane behavior changes, run focused Rust checks against the touched app-server mapper/handler tests plus any matching native `epiphany-core` smoke binary. The old Python smoke shims have been cut; do not resurrect them.
- For Epiphany control-plane behavior changes, run the focused existing library
or integration tests that exercise the changed consequence, plus touched
app-server mapper/handler tests when applicable. Do not create or resurrect a
standalone smoke executable merely to test a library or contract.
- Source, test, and target counts are pressure signals, not acceptance metrics.
Do not delete exact conflict, substitution, concurrency, lifecycle, re-entry,
or user-visible consequence proofs merely to improve a count. Preserve the
claim in the smallest owning library or integration surface, and delete a
verifier only when the claim is dead, duplicated, or covered by stronger
evidence.
- For Codex Rust work on this Windows machine, set `$env:CARGO_TARGET_DIR='C:\Users\Meta\.cargo-target-codex'`.
- Do not parallelize cargo builds or tests against the same target directory.

View file

@ -48,6 +48,24 @@ In the codex-rs folder where the rust code lives:
trivial; prefer new modules/files and keep `chatwidget.rs` focused on orchestration.
- When running Rust commands (e.g. `just fix` or `cargo test`) be patient with the command and never try to kill them using the PID. Rust lock can make the execution slow, this is expected.
## Source And Build Economy
- Verified behavior is the asset. Owned source, tests, crates, dependencies,
compiled targets, and build output are carrying costs. Prefer deletion,
collapse, or reuse before adding another mechanism, and minimize build fan-out
rather than chasing line-count reductions.
- Every new crate, binary, dependency, or test target must have a named owner,
live consumer, and invariant that an existing surface cannot coherently own.
A fixture or probe does not justify a shipped binary.
- Treat module-size guidance as a readability constraint, not an instruction to
manufacture crates or abstraction layers. Preserve consequence-bearing tests
and clear code when a smaller metric would make the machine harder to trust.
- Before a workspace-wide, all-target, or all-feature build, enumerate the
package and target scope, identify the output root, inspect its size and free
disk, and decide what may be retained. If a build is interrupted, inventory
and settle its output before retrying instead of accumulating a fresh target
directory.
Run `just fmt` (in `codex-rs` directory) automatically after you have finished making Rust code changes; do not ask for approval to run it. Additionally, run the tests:
1. Run the test for the specific project that was changed. For example, if changes were made in `codex-rs/tui`, run `cargo test -p codex-tui`.
@ -64,7 +82,10 @@ To that end: **resist adding code to codex-core**!
Particularly when introducing a new concept/feature/API, before adding to `codex-core`, consider whether:
- There is an existing crate other than `codex-core` that is an appropriate place for your new code to live.
- It is time to introduce a new crate to the Cargo workspace for your new functionality. Refactor existing code as necessary to make this happen.
- A new Cargo workspace crate owns a coherent boundary and repays its added
dependency graph, build target, and maintenance surface by removing or
isolating greater complexity. Do not introduce one merely to satisfy a module
line-count guideline.
Likewise, when reviewing code, do not hesitate to push back on PRs that would unnecessarily add code to `codex-core`.

View file

@ -23,6 +23,10 @@ harness, not an old language model product.
- Prefer end-to-end coherence over local cleverness. A small pipeline whose
parts visibly deserve to exist beats a large one full of compensators for
misunderstood complexity.
- Treat verified capability as the asset and owned source, tests, dependencies,
compiled targets, and build artifacts as carrying costs. Prefer deletion,
collapse, and reuse before adding another mechanism. Minimize conceptual
surface and build fan-out, not line count or readability.
- Use vivid language as an attention handle, not as mysticism. In Epiphany role
work, body means code structure, runtime shape, data flow, seams, and scars;
hands means bounded source actuation; soul means objective, invariants,
@ -151,6 +155,11 @@ Codex's original prompt carries useful operator scars. Keep them.
surrounding design.
- Follow local style and existing abstractions. Add new abstractions only when
they remove real complexity or match an established pattern.
- A new crate, executable target, dependency, service, schema, or persistent
type must have a named owner, live consumer, protected invariant, and a reason
the existing surface cannot coherently own it. Preserve consequence-bearing
verifier reach; deleting proof only to improve a size metric is not
simplification.
- Keep public APIs small, predictable, and easy to use.
- Do not add copyright or license headers unless asked.
- Add comments only when they save future readers real work.
@ -162,6 +171,11 @@ Codex's original prompt carries useful operator scars. Keep them.
- Start with focused checks that exercise the surface you changed, then broaden
as risk warrants.
- Before a workspace-wide, all-target, all-feature, or otherwise broad build,
enumerate its package and target scope, identify the output root, inspect
existing output and free disk, and decide the retention bound. If interrupted,
inventory and settle that output before retrying instead of routing around it
with another target or cache root.
- Treat proxy metrics, mocks, and narrow tests as suggestive, not conclusive.
Validate against the real objective when the real objective can be measured.
- Do not fix unrelated failures unless the user asks. Mention them when they