coilyco-flight-deck/agentic-os

Watch

doctrine: retire o2r agent-channel mandate from the public-safe Command delivery section (split from agentic-os-kai#677) #289

New issue

Closed

opened 2026-06-25 12:03:57 +00:00 by coilysiren · 6 comments

coilysiren commented

2026-06-25 12:03:57 +00:00

Owner

The June 2026 surface reduction archived otel-a2a-relay (and its -cli / -debug surfaces). The public-safe "Command delivery" section in AGENTS.md still mandates the o2r agent channel for agent-to-agent command delivery:

When the recipient is another agent, command delivery runs over an o2r agent channel (otel-a2a-relay, docs/agent-channel-requests.md) ... The concrete issuance and verification commands live in the kai-command-handover skill.

That now points a future agent at a dead surface, and the kai-command-handover skill it references has already been updated to mark o2r retired / human-mediated (agentic-os-kai#677).

Update the agentic-os/AGENTS.md "Command delivery" agent-recipient paragraph to replace the o2r mandate with human-mediated handover (route agent-to-agent through the human, no autonomous channel right now), keeping it public-safe. Mark o2r archived-not-deleted (revivable if autonomous agent handover is ever wired for real; coily-ops absorption tracked at ward #104).

Split out from agentic-os-kai#677 because that work landed in agentic-os-kai, while this paragraph lives in the public-safe agentic-os base. Filed by the container agent that carried #677 (it could not push to agentic-os from its scope).

The June 2026 surface reduction archived `otel-a2a-relay` (and its `-cli` / `-debug` surfaces). The public-safe **"Command delivery"** section in `AGENTS.md` still mandates the o2r agent channel for agent-to-agent command delivery: > When the recipient is another **agent**, command delivery runs over an o2r agent channel ([`otel-a2a-relay`](...), `docs/agent-channel-requests.md`) ... The concrete issuance and verification commands live in the `kai-command-handover` skill. That now points a future agent at a dead surface, and the `kai-command-handover` skill it references has already been updated to mark o2r retired / human-mediated (agentic-os-kai#677). Update the `agentic-os/AGENTS.md` "Command delivery" agent-recipient paragraph to replace the o2r mandate with human-mediated handover (route agent-to-agent through the human, no autonomous channel right now), keeping it public-safe. Mark o2r archived-not-deleted (revivable if autonomous agent handover is ever wired for real; coily-ops absorption tracked at ward #104). Split out from agentic-os-kai#677 because that work landed in `agentic-os-kai`, while this paragraph lives in the public-safe `agentic-os` base. Filed by the container agent that carried #677 (it could not push to `agentic-os` from its scope).

coilysiren added the

priority

autonomy

headless

labels

2026-07-01 04:29:18 +00:00

coilyco-ops commented

2026-07-01 04:39:13 +00:00

Owner

🔒 Reserved by ward agent --driver claude — container engineer-claude-agentic-os-289 on host kais-macbook-pro-2.local is carrying this issue (reserved 2026-07-01T04:39:09Z). Concurrent ward agent runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); --force overrides.

— Claude (she/her), via ward agent

🔒 Reserved by `ward agent --driver claude` — container `engineer-claude-agentic-os-289` on host `kais-macbook-pro-2.local` is carrying this issue (reserved 2026-07-01T04:39:09Z). Concurrent `ward agent` runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); `--force` overrides.  — Claude (she/her), via `ward agent`

coilysiren referenced this issue from a commit

2026-07-01 04:41:49 +00:00

docs: retire lingering o2r/agent-channel live references

coilysiren closed this issue

2026-07-01 04:41:49 +00:00

coilysiren commented

2026-07-01 04:42:07 +00:00

Author

Owner

WARD-OUTCOME: done - retired the last present-tense o2r/agent-channel references from the public-safe docs; the Command delivery mandate was already corrected on main.

The headline surprise: the actual ask - the AGENTS.md "Command delivery" agent-recipient paragraph - was already done. Commits baa719c and d0127bf had landed the human-mediated rewrite (o2r archived-not-deleted, ward#104 pointer) back on 2026-06-26, so this issue was effectively resolved but never closed. Rather than close on an empty change, I swept the repo for the loose ends the mandate fix left behind and found two docs still speaking of o2r in the present tense: dev-base-self-name.md listed "o2r channel traffic" as a live context for agent names, and dictatable-id-alphabet.md described the agent-channel protocol as if it still ran. Both now mark the channel archived with the ward#104 revival pointer, matching the archived-not-deleted spirit.

Nothing fought back - pre-commit's catalog suite passed first try, and the push fast-forwarded clean (the credential-lock warning was cosmetic, the ref moved). Confidence is high: these were public-safe markdown edits, low blast radius, and the doctrine surface is now internally consistent.

One deliberate non-change worth noting: the 2026-05-08 security-boundary finding under .agents/skills/ still names otel-a2a-relay, but that is a dated historical record and doctrine exempts those from legacy-reference cleanup, so I left it. No follow-ups needed.

WARD-OUTCOME: done - retired the last present-tense o2r/agent-channel references from the public-safe docs; the Command delivery mandate was already corrected on main. The headline surprise: the actual ask - the AGENTS.md "Command delivery" agent-recipient paragraph - was already done. Commits `baa719c` and `d0127bf` had landed the human-mediated rewrite (o2r archived-not-deleted, ward#104 pointer) back on 2026-06-26, so this issue was effectively resolved but never closed. Rather than close on an empty change, I swept the repo for the loose ends the mandate fix left behind and found two docs still speaking of o2r in the present tense: `dev-base-self-name.md` listed "o2r channel traffic" as a live context for agent names, and `dictatable-id-alphabet.md` described the `agent-channel` protocol as if it still ran. Both now mark the channel archived with the ward#104 revival pointer, matching the archived-not-deleted spirit. Nothing fought back - pre-commit's catalog suite passed first try, and the push fast-forwarded clean (the credential-lock warning was cosmetic, the ref moved). Confidence is high: these were public-safe markdown edits, low blast radius, and the doctrine surface is now internally consistent. One deliberate non-change worth noting: the 2026-05-08 security-boundary finding under `.agents/skills/` still names `otel-a2a-relay`, but that is a dated historical record and doctrine exempts those from legacy-reference cleanup, so I left it. No follow-ups needed.

coilyco-ops commented

2026-07-02 01:59:29 +00:00

Owner

🔒 Reserved by ward agent --driver claude — container engineer-claude-agentic-os-289 on host kais-macbook-pro-2.local is carrying this issue (reserved 2026-07-02T01:59:27Z). Concurrent ward agent runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); --force overrides.

— Claude (she/her), via ward agent

🔒 Reserved by `ward agent --driver claude` — container `engineer-claude-agentic-os-289` on host `kais-macbook-pro-2.local` is carrying this issue (reserved 2026-07-02T01:59:27Z). Concurrent `ward agent` runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); `--force` overrides.  — Claude (she/her), via `ward agent`

coilyco-ops commented

2026-07-02 01:59:33 +00:00

Owner

— Claude (she/her), via ward agent

🔒 Reserved by `ward agent --driver claude` — container `engineer-claude-agentic-os-289` on host `kais-macbook-pro-2.local` is carrying this issue (reserved 2026-07-02T01:59:27Z). Concurrent `ward agent` runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); `--force` overrides.  — Claude (she/her), via `ward agent`

coilyco-ops commented

2026-07-02 01:59:37 +00:00

Owner

— Claude (she/her), via ward agent

🔒 Reserved by `ward agent --driver claude` — container `engineer-claude-agentic-os-289` on host `kais-macbook-pro-2.local` is carrying this issue (reserved 2026-07-02T01:59:27Z). Concurrent `ward agent` runs are blocked until it finishes or the reservation goes stale (2h0m0s TTL); `--force` overrides.  — Claude (she/her), via `ward agent`

coilysiren commented

2026-07-02 02:01:23 +00:00

Author

Owner

WARD-OUTCOME: done - AGENTS.md Command-delivery paragraph already reflects human-mediated handover (o2r archived-not-deleted, ward#104 revival pointer); nothing left to change.

Honest retrospective: this one was already finished before I arrived. The substantive AGENTS.md rewrite landed in baa719c (Jun 26) - it replaced the o2r mandate with human-mediated-through-Kai, dropped the dead docs/agent-channel-requests.md link, kept the kai-command-handover skill pointer, and noted ward#104. d0127bf then split the section by execution model, and 9d9c24d (which carried closes #289) mopped up the last present-tense o2r references in two other docs. HEAD is even with origin/main and the issue was already CLOSED when this container spun up.

So there was no code to write - the right call was to verify rather than invent a change. I read the current paragraph, traced the three commits, and confirmed each of the issue's asks (human-mediated replacement, archived-not-deleted framing, revival tracking, public-safe, skill pointer) is satisfied on main. Confidence is high: the diff already exists and matches the spec exactly.

The only rough edge is process, not content: this looks like a duplicate/re-dispatch of an already-closed P1 - comments 6-8 are repeated coilyco-ops reservation notices as containers kept re-activating. Might be worth a look at why a closed issue with a landed closes-trailer still gets picked up for headless dispatch, but that's a ward-agent scheduling question, not anything owed to this issue.

WARD-OUTCOME: done - AGENTS.md Command-delivery paragraph already reflects human-mediated handover (o2r archived-not-deleted, ward#104 revival pointer); nothing left to change. Honest retrospective: this one was already finished before I arrived. The substantive AGENTS.md rewrite landed in baa719c (Jun 26) - it replaced the o2r mandate with human-mediated-through-Kai, dropped the dead docs/agent-channel-requests.md link, kept the kai-command-handover skill pointer, and noted ward#104. d0127bf then split the section by execution model, and 9d9c24d (which carried closes #289) mopped up the last present-tense o2r references in two other docs. HEAD is even with origin/main and the issue was already CLOSED when this container spun up. So there was no code to write - the right call was to verify rather than invent a change. I read the current paragraph, traced the three commits, and confirmed each of the issue's asks (human-mediated replacement, archived-not-deleted framing, revival tracking, public-safe, skill pointer) is satisfied on main. Confidence is high: the diff already exists and matches the spec exactly. The only rough edge is process, not content: this looks like a duplicate/re-dispatch of an already-closed P1 - comments 6-8 are repeated coilyco-ops reservation notices as containers kept re-activating. Might be worth a look at why a closed issue with a landed closes-trailer still gets picked up for headless dispatch, but that's a ward-agent scheduling question, not anything owed to this issue.

No milestone

No project

No assignees

2 participants

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

coilyco-flight-deck/agentic-os#289

No description provided.

Rows
Columns