Implement
Implement the accepted outcome. Repair ordinary known defects directly. Use the
existing intent; a clear edit needs no Plan first. An incoming report nobody
has accepted yet goes to Triage first. Implement owns source changes
and factual checks.
Rules that decide the result
- Fix the cause, not the oracle. Never loosen an assertion, tolerance,
golden, fixture or suppression, or substitute a mock or placeholder, to turn
a check green; nor change what the check runs against: code that acts
differently under test or CI, a newly silenced failure path, or a check made
non-blocking or skipped. A check that fails against accepted behavior points
at a product defect; changing what the check accepts is an acceptance change
and needs caller authority.
- Find live consumers before editing. Search the callers, readers, scripts,
docs and tests of every edited function, type, path, format or message. For
each one, state whether the change affects it and which check covers it. When
retiring or renaming, consumers include installations, lookups and old-name
invocations; preserve historical provenance. If the replaced interface never
merged or shipped and every consumer update fits the scope, update them and
delete the old version in the same change, with no alias or shim; otherwise
keep the old version working or report it.
- Name discriminating checks. A behavioral change preserves RED
for the expected missing behavior: name the check that fails before the edit
and passes after (Test's proof rule),
plus the check for each affected consumer.
- Report only what ran. Give exact commands and results. List every check
you did not or could not run as not run. Never call unrun work verified,
fixed or green. Output can show as soon as a command runs, so keep secrets
out when you build it: credentials stay in the environment, output that may
carry tokens goes to a file, and only redacted excerpts are shown.
When changing a public failure path, preserve its native failure signal and
check that its recovery advice matches any work already performed. For uncertain
write outcomes or unattended callers, use only the relevant checks in
Test's error recovery method.
For an accepted bounded shortcut, review feedback or a defect your change
introduced, an authorized commit, history rewrite or merge conflict, dependency
upgrade, removal, blocked operation, a decision read from another tool's state
or your own uncertain write, use only the matching change safety method.
These methods preserve existing authority; they grant no new operation.
For authorized service operations, load only the relevant procedure from
operations methods (reliability, delivery, incident
recovery, resilience, toil); ordinary edits owe no operations phase.
Workflow
- Read intent, acceptance, scope and repository boundaries before the first
write; reuse loaded contracts. RPI boundaries
apply when RPI is selected; caller-selected episode tracking follows the
session association reference.
Note the revision and files already modified, staged or untracked; report
only your own as changed, and ask the caller before editing one holding
others' edits.
- Carry the accepted behavior examples forward unchanged. Use repository
domain names in symbols and tests; check observable outcomes through the
relevant interface. Run the smallest applicable check before and after
editing. Pure refactors, relocations or docs may have an honest green
baseline. Prefer existing tests or small discriminating probes.
- Make the smallest in-scope change. When repairing discovery or checks,
preserve the consumer's existing input selection; fixing an error path does
not authorize a wider scan. Use a negative control when exclusion matters.
Check a representative change against existing constraints before bulk
propagation, and the authored source set before broad regeneration. A
scripted edit across files first confirms that each anchor matches exactly
once; on zero or several matches, write nothing and re-derive it. Repair
known failures directly and verify the exact result with a check; a repair
does not start another review. A disproved assumption may change the
approach within scope; use Plan only for consequential uncertainty.
- Read the repository's actual check recipe, including instrumentation and
environment, rather than reconstructing it from memory. Run targeted tests
and lint/static checks before broad integration, and required full checks at
integration. Reuse exact-input receipts only while source, tool and relevant
environment match. Neither bypass required hooks nor replay a check just to
rename its receipt. A required CI job's known failure is actionable before
the run ends: repair it within scope, keep the failed subject's evidence and
rerun affected checks. Pending jobs do not imply success. A new check, hook
or lint rule counts once the aggregate command or CI runs it (a Git hook needs
an install path;
.git/hooks is not cloned) and it fails on a deliberate
violation; else list it under gaps as not wired. Use a local server already
on a port only if it provably serves this checkout; stop only processes you started.
- Refactor while acceptance remains green. Inspect changed tests, fixtures,
goldens, tolerances, suppressions and specification text against original
intent before handoff. Reread that intent at its source, not a compacted
summary, and check the whole delta against it, untracked files included.
Remove leftovers of abandoned approaches, temporary probes (one shared
marker lets a single search prove none remain) and files your checks made;
list any path you cannot explain under
gaps as a scope amendment.
- Have the runtime derive actual changed paths and content identity: read
changed paths from Git (
git status --porcelain,
git diff --name-only <base>), never from memory;
ao provenance manifest --root <repo> with one --include <path> per
declared path digests only the paths you declare and runs no Git. A delegated
increment awaiting integration returns an exact commit or runtime-derived
content digests, author context ID and check facts in the existing handoff.
Copy that ID from the session, subagent or thread ID the runtime exposes, or
label a caller-supplied one as supplied; otherwise report it as unknown,
never made up. The integrating caller derives subject-manifest.v1 (AgentOps
schema subject-manifest.v1.schema.json, AgentOps source checkout only) over the complete final
subject before judgment; an independently judged increment needs its own
manifest. Do not generate both merely because work was delegated. Where the
repository records AgentOps evidence bindings, also run the advisory
ao provenance evidence-orphans --root <repo-root> with one --changed <path>
per derived path. Disclose the rows whose cause is changed_path or both
(they bind your paths); count the rest under gaps as pre-existing drift,
and refresh a binding only when its owner asks. Without ao, list it under
not run.
- Return the handoff below.
Handoff
Keep full logs at their source; do not duplicate inventories or status documents.
changed: <runtime-derived paths>; identity: <commit or content digests>
author: <runtime session/subagent/thread ID, "ID (supplied)" or "unknown">; when an integrator or fresh read follows
checks: <exact command> -> <result>, one per line; before and after for a behavioral change
not run: <check> -> <reason>; "none" only when every relevant check ran
consumers: <file:line> -> <covering check, or "uncovered">
not_as_briefed: <steps refused, held, worked around or unverified, one line with its reason each> | none
gaps: <missing or truncated evidence, known failures, scope amendments needed> | none
cause: <defect repairs only: trigger -> mechanism -> symptom at file:line; sites left as follow-up>
Report an uncovered live consumer for a caller scope amendment and continue
independent authorized work. Disclose a reversible implementation choice only
when its alternative has a material consequence the caller may want to change;
name the choice, alternative and reason in the handoff, without a fixed worksheet.
Generated companions already included as scope require no new approval; a change
to accepted direction, behavior or acceptance always needs caller authority.
Diagnosis, scaffolding and delegated work
Before repairing a failure you cannot yet explain, read and follow the
diagnosis loop. The same
file holds the experiments for races, leaked state, hangs, regressions and
failures that cross components.
When scaffolding is the requested change, start from the repository's existing
layout (or the ecosystem's standard one for a new project) and one working
vertical slice; avoid placeholder success paths and a new framework for a
one-off operation.
Prefer current-session execution. When delegation is authorized, or the caller
asks for a one-shot adapter or a factory, read
delegated work before dispatching or writing it.
Finish
Specialists advise only; known defects stay implementation work. A genuine
causal stall admits one bounded fresh helper under RPI's
stall rule; retries renew no bound.
Return facts, not semantic PASS. An implement-only handoff does not authorize
Git, tracker or delivery transitions; existing caller authority remains usable.
A full outcome request finishes on its checks and CI; it continues to one fresh
independent judgment only on Validate's triggers.
1---2name: implement-53description: 'Make the code change a ticket or bug report asks for, then prove it with the repo''s own checks. Use when: told to fix, change or build something, such as "fix this bug".'4practices:5- tdd6- refactoring7- small-batch-flow8hexagonal_role: driving-adapter9consumes: []10produces:11- subject-manifest.v112output_contract: 'content identity, author context ID and check facts through the native handoff; subject-manifest.v1 at the judgment boundary'13context_rel:14- kind: customer-of15 with: plan16skill_api_version: 117user-invocable: true18metadata:19 source: https://skillmd.com/skills/hashgraph-online/implement-520 graph_root: true21 tier: execution22 dependencies: []23 capabilities: [execute_one_experiment, collect_factual_evidence]24 effects: [modify_declared_subject, derive_subject_manifest]25 canonical_status: canonical26 disposition: keep27 triggers: ["execute the next wave", "per-lane evidence", "fix this bug", "CI is red", "loosen the assertion"]28---2930# Implement3132Implement the accepted outcome. Repair ordinary known defects directly. Use the33existing intent; a clear edit needs no Plan first. An incoming report nobody34has accepted yet goes to [Triage](../triage/SKILL.md) first. Implement owns source changes35and factual checks.3637## Rules that decide the result3839- **Fix the cause, not the oracle.** Never loosen an assertion, tolerance,40 golden, fixture or suppression, or substitute a mock or placeholder, to turn41 a check green; nor change what the check runs against: code that acts42 differently under test or CI, a newly silenced failure path, or a check made43 non-blocking or skipped. A check that fails against accepted behavior points44 at a product defect; changing what the check accepts is an acceptance change45 and needs caller authority.46- **Find live consumers before editing.** Search the callers, readers, scripts,47 docs and tests of every edited function, type, path, format or message. For48 each one, state whether the change affects it and which check covers it. When49 retiring or renaming, consumers include installations, lookups and old-name50 invocations; preserve historical provenance. If the replaced interface never51 merged or shipped and every consumer update fits the scope, update them and52 delete the old version in the same change, with no alias or shim; otherwise53 keep the old version working or report it.54- **Name discriminating checks.** A behavioral change preserves RED55 for the expected missing behavior: name the check that fails before the edit56 and passes after ([Test's proof rule](../test/SKILL.md#critical-constraints)),57 plus the check for each affected consumer.58- **Report only what ran.** Give exact commands and results. List every check59 you did not or could not run as not run. Never call unrun work verified,60 fixed or green. Output can show as soon as a command runs, so keep secrets61 out when you build it: credentials stay in the environment, output that may62 carry tokens goes to a file, and only redacted excerpts are shown.6364When changing a public failure path, preserve its native failure signal and65check that its recovery advice matches any work already performed. For uncertain66write outcomes or unattended callers, use only the relevant checks in67[Test's error recovery method](../test/references/error-recovery.md).6869For an accepted bounded shortcut, review feedback or a defect your change70introduced, an authorized commit, history rewrite or merge conflict, dependency71upgrade, removal, blocked operation, a decision read from another tool's state72or your own uncertain write, use only the matching [change safety method](references/change-safety.md).73These methods preserve existing authority; they grant no new operation.7475For authorized service operations, load only the relevant procedure from76[operations methods](references/operations.md) (reliability, delivery, incident77recovery, resilience, toil); ordinary edits owe no operations phase.7879## Workflow80811. Read intent, acceptance, scope and repository boundaries before the first82 write; reuse loaded contracts. [RPI boundaries](../rpi/references/boundaries.md)83 apply when RPI is selected; caller-selected episode tracking follows the84 [session association reference](../orchestrate/references/session-associations.md#work-to-session-associations).85 Note the revision and files already modified, staged or untracked; report86 only your own as changed, and ask the caller before editing one holding87 others' edits.882. Carry the accepted behavior examples forward unchanged. Use repository89 domain names in symbols and tests; check observable outcomes through the90 relevant interface. Run the smallest applicable check before and after91 editing. Pure refactors, relocations or docs may have an honest green92 baseline. Prefer existing tests or small discriminating probes.933. Make the smallest in-scope change. When repairing discovery or checks,94 preserve the consumer's existing input selection; fixing an error path does95 not authorize a wider scan. Use a negative control when exclusion matters.96 Check a representative change against existing constraints before bulk97 propagation, and the authored source set before broad regeneration. A98 scripted edit across files first confirms that each anchor matches exactly99 once; on zero or several matches, write nothing and re-derive it. Repair100 known failures directly and verify the exact result with a check; a repair101 does not start another review. A disproved assumption may change the102 approach within scope; use Plan only for consequential uncertainty.1034. Read the repository's actual check recipe, including instrumentation and104 environment, rather than reconstructing it from memory. Run targeted tests105 and lint/static checks before broad integration, and required full checks at106 integration. Reuse exact-input receipts only while source, tool and relevant107 environment match. Neither bypass required hooks nor replay a check just to108 rename its receipt. A required CI job's known failure is actionable before109 the run ends: repair it within scope, keep the failed subject's evidence and110 rerun affected checks. Pending jobs do not imply success. A new check, hook111 or lint rule counts once the aggregate command or CI runs it (a Git hook needs112 an install path; `.git/hooks` is not cloned) and it fails on a deliberate113 violation; else list it under `gaps` as not wired. Use a local server already114 on a port only if it provably serves this checkout; stop only processes you started.1155. Refactor while acceptance remains green. Inspect changed tests, fixtures,116 goldens, tolerances, suppressions and specification text against original117 intent before handoff. Reread that intent at its source, not a compacted118 summary, and check the whole delta against it, untracked files included.119 Remove leftovers of abandoned approaches, temporary probes (one shared120 marker lets a single search prove none remain) and files your checks made;121 list any path you cannot explain under `gaps` as a scope amendment.1226. Have the runtime derive actual changed paths and content identity: read123 changed paths from Git (`git status --porcelain`,124 `git diff --name-only <base>`), never from memory;125 `ao provenance manifest --root <repo>` with one `--include <path>` per126 declared path digests only the paths you declare and runs no Git. A delegated127 increment awaiting integration returns an exact commit or runtime-derived128 content digests, author context ID and check facts in the existing handoff.129 Copy that ID from the session, subagent or thread ID the runtime exposes, or130 label a caller-supplied one as supplied; otherwise report it as unknown,131 never made up. The integrating caller derives `subject-manifest.v1` (AgentOps132 schema [subject-manifest.v1.schema.json](https://github.com/boshu2/agentops/blob/main/schemas/subject-manifest.v1.schema.json), AgentOps source checkout only) over the complete final133 subject before judgment; an independently judged increment needs its own134 manifest. Do not generate both merely because work was delegated. Where the135 repository records AgentOps evidence bindings, also run the advisory136 `ao provenance evidence-orphans --root <repo-root>` with one `--changed <path>`137 per derived path. Disclose the rows whose cause is `changed_path` or `both`138 (they bind your paths); count the rest under `gaps` as pre-existing drift,139 and refresh a binding only when its owner asks. Without `ao`, list it under140 `not run`.1417. Return the handoff below.142143## Handoff144145Keep full logs at their source; do not duplicate inventories or status documents.146147```text148changed: <runtime-derived paths>; identity: <commit or content digests>149author: <runtime session/subagent/thread ID, "ID (supplied)" or "unknown">; when an integrator or fresh read follows150checks: <exact command> -> <result>, one per line; before and after for a behavioral change151not run: <check> -> <reason>; "none" only when every relevant check ran152consumers: <file:line> -> <covering check, or "uncovered">153not_as_briefed: <steps refused, held, worked around or unverified, one line with its reason each> | none154gaps: <missing or truncated evidence, known failures, scope amendments needed> | none155cause: <defect repairs only: trigger -> mechanism -> symptom at file:line; sites left as follow-up>156```157158Report an uncovered live consumer for a caller scope amendment and continue159independent authorized work. Disclose a reversible implementation choice only160when its alternative has a material consequence the caller may want to change;161name the choice, alternative and reason in the handoff, without a fixed worksheet.162Generated companions already included as scope require no new approval; a change163to accepted direction, behavior or acceptance always needs caller authority.164165## Diagnosis, scaffolding and delegated work166167Before repairing a failure you cannot yet explain, read and follow the168[diagnosis loop](references/difficult-failures.md#diagnosis-loop). The same169file holds the experiments for races, leaked state, hangs, regressions and170failures that cross components.171172When scaffolding is the requested change, start from the repository's existing173layout (or the ecosystem's standard one for a new project) and one working174vertical slice; avoid placeholder success paths and a new framework for a175one-off operation.176177Prefer current-session execution. When delegation is authorized, or the caller178asks for a one-shot adapter or a factory, read179[delegated work](references/delegated-work.md) before dispatching or writing it.180181## Finish182183Specialists advise only; known defects stay implementation work. A genuine184causal stall admits one bounded fresh helper under RPI's185[stall rule](../rpi/SKILL.md#causal-stall-and-bounds); retries renew no bound.186187Return facts, not semantic PASS. An implement-only handoff does not authorize188Git, tracker or delivery transitions; existing caller authority remains usable.189A full outcome request finishes on its checks and CI; it continues to one fresh190independent judgment only on [Validate's triggers](../validate/SKILL.md#when-a-fresh-judgment-is-worth-it).