Returns raw quota + the full conversion params (quota_per_unit, quota_display_type,
usd_exchange_rate) AND server-computed display_quota / display_used_quota, so the
desktop client renders correctly in USD / CNY / tokens without a ×500000 hack or a
missing-exchange-rate bug. Contract of record (server applies it):
USD : raw / quota_per_unit
CNY : raw / quota_per_unit * usd_exchange_rate
tokens : raw (no division)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Verified against code.claude.com/docs/en/permission-modes and fixed 6 points:
1. acceptEdits: add PowerShell tool auto-approvals (Set-Content/Add-Content/
Clear-Content/Remove-Item + aliases) and env-prefix/process-wrapper note.
2. Protected paths: .claude exception is ONLY .claude/worktrees (was wrongly
widened to commands/agents/skills); add per-mode protected-write table.
3. defaultMode:"auto" ignored from project files since v2.1.142+.
4. dontAsk: read-only Bash commands also run without allow rules; explicit ask
rules are denied (not prompted).
5. auto conversational boundary: stays in force until user lifts it; Claude's own
judgment doesn't lift it; lost on context compaction; use deny rule for hard.
6. auto consecutive-failure fallback: 3-in-a-row / 20-total pauses & re-prompts;
-p non-interactive aborts the session. Plus dropped broad allow-rules on entry.
Co-authored-by: chenchen <chenchen@xinghanlab.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Per Fasthei's CHANGES_REQUESTED on #15: deleting all sub/swarm docs left no
tracking entry for the still-in-development swarm capability. Adds
docs/integration/heicode-swarm-deferred.md stating HM does not implement swarm
runtime, the old HM-orchestrated sub/swarm model is deprecated, new swarm work
lives on AM/Swarm side, and a per-file migration mapping (deprecated / deferred
/ superseded) for all 11 deleted docs. Links it from docs/README.md so deletion
no longer drops context.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds a reading guide, the full 6 permission modes (incl. dontAsk) + auto-mode
admin enablement toggle, confirmed role facts (UsageView roles; full matrix is
external; iam pages 404), deep permission-rule syntax (Bash spacing, Read/Edit
anchors, MCP/Agent), real MCP credential mechanisms (headersHelper/${VAR}/OAuth),
expanded usage/cost/attribution/analytics, plus two big appendices: verbatim
config examples (A1–A10) and step-by-step end-to-end flows (managed-settings
lifecycle, MCP allow/deny worked example, auto-mode force-push decision, auth
credential selection). Sourced from re-fetching the 6 core docs + permission-
modes/permissions/mcp/costs/monitoring-usage/analytics.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Aligns HM's capabilities to three tiers mirroring Claude's proven split, grounded
in HM's existing primitives. Defines positioning + billing per tier, a capability
× tier matrix (use layer = Individual; manage/allocate = Teams; compliance/hard-
enforcement/private = Enterprise) with HM code landing points, the Teams-hook vs
Enterprise-moat rationale, HM's differentiators over Claude (gateway-side hard
cost control, signed per-org policy push, server-side hard enforcement, native
agent catalog), a P0–P3 rollout mapping, and the open pricing decisions to settle.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Expands §1.2 into a proper three-tier section: positioning per tier, a
capability × tier table (with source + whether the doc is explicit), the
billing differences, and the documented contradiction (authentication lists
"managed policy settings" as Enterprise-only while the server-managed-settings
feature page says Teams+ can use it). Also flags that most controls hinge on
provider (Anthropic-direct + admin write access) rather than subscription tier.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Standalone reading reference faithfully consolidating the six official Claude
docs (admin-setup, authentication, server-managed-settings, settings,
managed-mcp, auto-mode-config): auth methods + precedence, roles/seats/admin,
settings scope precedence & merge rules, server-managed-settings mechanism,
full managed-only/permissions/model/sandbox/MCP key reference, managed-mcp
seven modes + allow/deny evaluation, auto-mode classifier, usage/audit, and
quick-reference tables. No HM design — pure Claude reference for study.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backend:
- loadAgentTemplate now requires status='active' — a known template_key can no
longer deploy a template an admin deactivated (matches the client list).
- refreshAgentStatus no longer lets AM's eventually-consistent live status
resurrect a user-initiated "stopped" agent.
- HeicodeStopAgent persists via field-level Updates (not a stale full-row Save),
matching refreshAgentStatus discipline.
- Drop dead amStartResult.AccessToken field (AM's token is never used; HM mints
its own per-agent token).
Frontend:
- deploy-agent statusLabel: add the missing pending/starting → 启动中 branch so a
just-deployed agent isn't shown as raw English fallback.
- cockpit 最近部署: map template_id → Chinese template name (consistent with the
deploy/status pages) instead of showing the raw key.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Token IS transmitted by HM (confirmed); the agent didn't enforce it because AM
hadn't deployed the image containing the §5 check to production. So the debug
log is unnecessary — removed. Contract §0.1 updated: token-check is "code-ready,
pending AM prod image", not a HM gap. UI access-token/direct-URL display kept.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- runtime: log the env KEY NAMES (never values) sent to AM on start, so we can
confirm AGENT_ACCESS_TOKEN is actually transmitted without leaking secrets.
- web/运行状态: detail panel now shows the full direct-connect URL (copyable) and
the per-agent access token (masked + reveal + copy) with a hint that the client
sends it as X-Agent-Access-Token. These are what's actually needed to hand the
agent to a client / test it — previously only the bare hostname was shown.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
AM shipped its updated runtime + fixes; live re-test confirms deploy → running
→ direct /health & /message/send (task completed) → stop → delete all pass, and
HM's per-agent access_token is minted + returned (non-empty UUID). Two AM-side
gaps remain: (1) the running agent does NOT enforce X-Agent-Access-Token
(no-token request still 200, agent card auth_required=None) — token isolation
not actually active; (2) subdomain is http:// (token + api_key in cleartext).
Contract §0.1 and client doc status notes updated accordingly.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- AM contract: fix stale lifecycle path defaults in the env-override table
(/agents/{agent_id}, /agents/{agent_id}/stop — matches code, not the old
/api/agent/... values); correct the self-check create line to POST /agents;
align the verify-endpoint example to the real production response shape
(user_id is a string, agent_id included, miss returns {valid:false}).
- Client API: §0 overview now states HM mints the per-agent access_token
(AM no longer "returns" it).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
amStartTemplateAgent no longer claims OPENAI_API_KEY is uninjected or that
access_token is left empty for a future V2 path: the handler now mints +
injects both OPENAI_API_KEY and the per-agent AGENT_ACCESS_TOKEN (forwarded
as-is), and HM returns its own access_token to the client.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Per the chosen design, the agent authorizes callers by comparing the request
header X-Agent-Access-Token against its env AGENT_ACCESS_TOKEN (constant-time),
no HM round-trip. AM contract §3.1 now states ① as the agreed integration with
Python pseudo-code; the /agent-access/verify endpoint is demoted to an optional
fallback. Client API §6 spells out the client's job: send X-Agent-Access-Token
on every direct-connect request.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
HM now mints a random per-agent access token at deploy, injects it into the
agent env (AGENT_ACCESS_TOKEN + HEICODE_AGENT_ID) and returns it to the
deploying client (agent list access_token). Only the owning user receives it,
so only they can drive the agent — closing the gap where any valid sk- could
drive any agent and exfiltrate its mounted resources.
AM authorizes the caller either locally (compare to its env token) or via the
new public POST /api/heicode/agent-access/verify {agent_id, access_token} ->
{valid, user_id} (constant-time compare, no info leak on miss). AM may opt out.
Docs: AM contract §3.1 + client API §6 updated; access_token no longer empty.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
1) Deploy Sub Agent: stop hiding resource types AM doesn't support yet (vm) —
show them but disabled with a "暂不支持" badge + tooltip, so users see their
bindings instead of wondering where the vm went.
2) Overview (概述/cockpit): replaced the old task-model status cards
(Running/Completed/Failed/裁决 + "code delivery runs") with template-agent
semantics — Agent 总数 / 运行中 / 启动中 / 异常, "最近部署 · 我的 Agent",
all Chinese. Dropped unused t()/useTranslation/PlayCircle.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Re-add a run-status menu (I shouldn't have removed it) but showing ONLY what the
template-agent model actually has — no old 待确认/sub_agile/子任务流/智能体任务图/
产物/SK快照/合并时间线. New /agent-status page:
- summary counts (total / running / starting / other)
- agent list (template Chinese name, live status badge, subdomain, #resources)
- detail panel: live status (polls /agents/{id}/status from AM), copyable
subdomain, mounted resources (names from /api/resources), runtime_id, timestamps.
Sidebar item "运行状态" + zh key added; build + tsc clean; route registered.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- new §0.1 联调结果: real production test outcomes — create POST /agents works
(returns access_info.domain/namespace), DELETE /agents/{id} 500s (AM
UnboundLocalError bug), POST /agents/{id}/stop 404 (no endpoint), agent stays
Pending / subdomain unreachable. The 3 AM-side blockers listed up top.
- OPENAI_API_KEY is now injected (a minted user new-api sk-, billed to the user,
revoked on delete; verified working at /v1) — §1.1 env + §2 updated.
- §3 client<->agent: now documents AM's A2A protocol (message/send · stream),
api_key auth, and flags the per-user isolation security gap.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The old /deployments (运行总览) + /tasks/$id pages rendered the new template agents
with meaningless old sub-task fields (待确认/sub_agile/子任务流/智能体任务图/产物/
SK快照/合并时间线 — all empty). Removed the whole old-model UI cluster and pointed
the Overview dashboard at the real new endpoint:
- deleted features/agent-console, features/agent-hub, features/tasks and the
routes /deployments and /tasks/$id.
- cockpit (Overview) now lists deployed template agents from /api/heicode/agents
(status counts + recent agents), not the old listAgentDeployments.
- removed the "运行总览" sidebar item + dead /deployments links in footer /
top-nav / sidebar-config.
- routeTree regenerated; frontend build + tsc clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ran the full desktop-client flow against production (real account) and updated
the doc to the verified responses:
- capabilities model = gpt-5.4; modes still returned but ignored.
- agent object: access_token is "" (AM issues none — connect via A2A api_key),
subdomain is a host (dep-xxx.taijiagnet.com), status starts Pending; real
binding_ids; shown as "生产实测" examples.
- §4: clarified the template `model` field is a role-tier hint (opus), not the
runtime model (gpt-5.4 injected at deploy).
- §5: stop currently fails / delete returns runtime_cleanup:"failed" (AM stop
missing + delete bug); HM removes the local record regardless.
- §6: must wait for status=running / health 200 before connecting; noted the
agent currently stays Pending and the subdomain isn't reachable yet (AM side).
- top banner: per-endpoint production verification note.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Deploy now also passes OPENAI_API_KEY (the 3rd of OPENAI_BASE_URL/MODEL_NAME/
OPENAI_API_KEY AM expects). HM mints a hidden, unlimited-quota sk- token for the
user per agent ("sk-"+key, billed to the user), injects it as OPENAI_API_KEY, and
stores the token id on the deployment. The token is revoked on delete and rolled
back if AM start / persist fails (no leaked keys). gateway accepts Bearer
sk-<key> (middleware strips sk-).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Preset templates carry a Claude-style frontmatter model (opus/sonnet) which is
NOT a model on the HM gateway. Passing it as MODEL_NAME would make the agent's
model calls fail. Now MODEL_NAME defaults to the gateway model gpt-5.4
(env AGENT_RUNTIME_DEFAULT_MODEL); a non-Claude-tier template model is honored.
Tests + AM contract doc updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
POST /api/devices/logout (UserOrV2DeviceAuth): a device-signed client revokes its
OWN bound token via the signed X-Heicode-Device-Id (cannot touch other devices);
a session/JWT caller may pass {device_id}. Idempotent. The existing DELETE
/api/devices/:id revoke is session-only, so device clients had no self-logout —
this closes that gap. Documented in the client API doc §1.1.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real test showed AM's DELETE /agents/{id} currently 500s (AM bug), which left the
HM record permanently undeletable. HeicodeDeleteAgent now treats the AM delete as
best-effort: on AM failure it logs + still removes the local record and returns
runtime_cleanup:"failed" so the user can always clear their list. (Stop stays
strict — AM has no /stop endpoint yet, pending their confirmation.)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real production test (user account) confirmed HM->AM POST /agents creates a real
agent (returned subdomain + status). But stop/delete still used the old
/api/agent/agents/{id} defaults and 404'd. Aligned status/stop/delete defaults to
the same namespace as create: /agents/{id}, /agents/{id}/stop. AM contract doc
notes these are HM's best guess pending AM's confirmation of the real lifecycle
endpoints (their doc only specified create).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adapt HM's template-agent integration to AM's actual CODING_A2A API (per their
doc), keeping it isolated in agent_template_runtime.go:
- start payload -> AM's POST /agents { name, template:"coding_a2a_agent",
framework:"A2A", config:{user_id,...}, env } with the template .md folded into
env.AGENT_INSTRUCTION_TEXT, template_key -> AGENT_ROLE_NAME, model gateway via
OPENAI_BASE_URL + MODEL_NAME (OPENAI_API_KEY left to the client per A2A request).
- response parse -> access_info.domain/external_ip -> subdomain, namespace/name
-> runtime_id; AM issues no access_token (client uses A2A api_key).
- env names aligned to AM: GIT_DEFAULT_BRANCH, POSTGRES_* (was PG_*),
AZURE_BLOB_ACCOUNT_NAME/CONTAINER/ACCOUNT_KEY (was BLOB_*); source keys aligned
to the resource-binding form (db_name/username/database_password/access_key).
Only AM-supported types (git/mysql/postgres/azure-blob); vm/redis/mongo/bucket
now rejected as unsupported until AM adds them.
- frontend: resources page splits DB into MySQL/PostgreSQL (correct provider),
drops vm; deploy page hides unsupported resource types.
- docs: AM contract + client doc updated to the real env names, payload, and the
A2A direct-connect (message/send · message/stream) + api_key auth.
- tests updated for the new env names + AM payload/response shape. All green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Hand-off doc for the agent_management team: the 4 endpoints AM must implement
(start/status/stop/delete) with exact request/response (grounded in the isolated
adapter agent_template_runtime.go), the env naming convention AM templates must
read (git/vm/db/blob/bucket), AM's responsibilities (inject .env, validate the
agent access token, models via HM /v1/*, no secret leakage), the client<->agent
direct SSE contract AM needs to define, security requirements, the env-overridable
paths, and a joint integration checklist.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Closes the last review finding: the AM-issued access_token was stored in plaintext
in agent_deployments. Now sealed with common.SealWithCryptoSecret on write and
unsealed on read — the client still receives the plaintext token (no contract
change), but a DB dump no longer exposes live agent tokens. Legacy/plaintext
values and an unset CryptoSecret degrade gracefully (pass-through). Round-trip test added.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two review agents (backend + frontend) audited the template-agent feature. Real
findings fixed (verified the one false-positive — okOrThrow already reads the
top-level success because the response interceptor returns response, not
response.data):
Backend:
- buildAgentEnvFromBindings now filters status='active' so a revoked/disabled
binding can't still have its KV secret resolved + injected.
- same-type binding env collision (e.g. two git repos -> GIT_REPO_URL) now errors
instead of silently overwriting.
- refreshAgentStatus uses a field-level Updates (status/updated_at) instead of
Save(full row) to avoid clobbering a concurrent stop/delete.
- tests added: same-type collision rejected, inactive binding rejected.
Frontend:
- deploy/stop/delete now pass noBusinessError so the global interceptor stops
double-toasting the raw English backend message; okOrThrow+onError give one
friendly Chinese error. Extracted okOrThrow/noBusinessError into lib/api.ts.
- resources page revoke + create + KV-secret-write now check success (okOrThrow)
so a failed unbind / credential write no longer falsely reports success.
- clipboard copy wrapped in try/catch (no false "copied" on failure).
- resources group label shows Chinese resource-type names.
Go + frontend builds/tests green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Review of the client doc against the real code found and fixed:
- §1 auth was not self-contained (deferred the canonical to the deprecated doc).
Inlined the full signing contract verified against middleware/device_signature.go:
the exact header set, the fixed-order canonical string (method/path/ts/nonce/
fingerprint/eph_pubkey/sha256(body)), ed25519(sha256(canonical)), the heicode-aead-v1
encrypted-body rules, and the X-Heicode-Auth-Error / X-Heicode-Server-Time failure
headers.
- §2 auth mismatch (accuracy bug): /api/user/self is UserAuth (session/JWT), NOT
device-signed — a device-only client cannot call it. Marked it optional and
clarified the two different auth schemes (/api/user/self* vs /api/heicode/*).
- §8: documented that failures return HTTP 200 with success:false (client MUST
read success), and that error.retryable is always false (decide retry by code).
- §10 inventory: corrected /api/user/self auth + added /self/models.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The 普通 sub task-orchestration model was replaced by the template-agent model
and its backend deleted. Removed the now-obsolete docs describing it:
- heicode-desktop-sub-agile-api.md, heicode-desktop-subagile-e2e-demo.md
- heicode-desktop-unified-api.md, heicode-sub-mode-flow-spec.md
- 普通sub敏捷模式-AgentManager对接任务清单.md
- AgentManager普通sub{产物回调缺失问题,剩余补充要求,联调整改要求}.md
Fixed dangling references in the new docs (client-api / template-agent-model).
Swarm (蜂群) docs kept — different mode, out of scope.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
chenchenandClaude Opus 4.8