akiiwa-games/AgentCompass / skills/agent-compass
agent-compass
FlaggedChoose owner, model and effort, outbound authorization, advisor use, and continuation for delegated work. Reads local usage and effective account policy, profiles the task with the configured System-One scorer, then allocates a compatible worker. Use for independent worker or advisor work; ordinary local execution needs no routing.
scripts/released_names.py
View on GitHub"""Model names a released version resolved through an alias that a later release removed or re-targeted, and the row
that release resolved each to: what the user meant by the name when they wrote it into local state (AGENTS.md: what a
released version wrote for the user is kept).
`V0_7_0` is the diff of the `aliases` tables of `references/accounts.json` between the `v0.7.0` tag and this release:
every name v0.7.0 resolved through an alias to one of its rows that this release resolves otherwise (the alias removed,
the name now a row of its own, or the alias re-targeted). It is read by the upgrade steps of the local files keyed by a
model name (`user_scores.LOCAL_UPGRADES`, schema 3 to 4) and where an auto bundle's v0.7.0 id is replayed
(`auto_bundles.derive`, `as_v0_7_0`).
test_released_names.py checks it against the tag.
"""
V0_7_0 = {
'claude-fable-5': 'claude-fable-5-2', # now a row of its own
'claude-opus-5': 'claude-opus-5-5', # now a row of its own
'claude-sonnet-5': 'claude-sonnet-5-5', # now a row of its own
'deepseek-flash': 'deepseek-v4.1-flash-free', # re-targeted to deepseek-v4.1-flash-1017
'deepseek-v4-flash': 'deepseek-v4.1-flash-free', # now a row of its own
'deepseek-v4.1-flash': 'deepseek-v4.1-flash-free', # now a row of its own
'gemini-4-pro': 'gemini-4-argon-high', # alias removed
'gemini-4-ultra': 'gemini-4-argon-high', # alias removed
'glm-5.3-flash': 'glm-5.3-flash-free', # now a row of its own
'grok-4': 'grok-4.7', # alias removed
'kimi-k3.1-pro-highspeed': 'kimi-k3.1-pro', # now a row of its own
'mimo-v2.6-pro-ultraspeed': 'mimo-v2.6-pro', # now a row of its own
}
def v0_7_0_row(name):
"""The row v0.7.0 resolved `name` to where this release resolves it otherwise, else None. v0.7.0
(user_scores.resolve_model_key) read a name as a row, then an alias, then its last `/` segment as a row, then that
segment as an alias; a name or segment in `V0_7_0` is one it resolved through an alias this release changed. No
provider-qualified alias of v0.7.0 has a last segment in `V0_7_0` (test_released_names.py), so the segment rule is
v0.7.0's exactly."""
leaf = str(name).rsplit('/', 1)[-1]
return V0_7_0.get(name) or V0_7_0.get(leaf)
def as_v0_7_0(accounts):
"""`accounts` (accounts.json) with every name of `V0_7_0` resolving as v0.7.0 resolved it: an alias of the row it
meant, not a row or an alias of its own. A name resolved through it (auto_bundles.default_profile: the advertised
spelling, its last segment, case, effort and vendor suffixes) reaches the row v0.7.0 reached; read where the id
v0.7.0 gave an auto bundle is replayed (auto_bundles.derive). v0.7.0 and v0.7.1 declared no row of a served name
(`provider_names`, added after v0.7.1), so the replay declares none: the id v0.7.1 gave a bundle whose name a
provider or its maker now declares is the id v0.7.0 gave it."""
models = {k: v for k, v in ((accounts or {}).get('models') or {}).items() if k not in V0_7_0}
aliases = {**{k: v for k, v in ((accounts or {}).get('aliases') or {}).items() if k not in V0_7_0}, **V0_7_0}
return {**(accounts or {}), 'models': models, 'aliases': aliases,
'provider_names': {'statement': 'v0.7.0 and v0.7.1 declared none', 'defaults': {}, 'rows': {}}}
def current_key(name, accounts):
"""The key a name resolves to under `accounts` (accounts.json) the way an auto bundle's id is keyed
(auto_bundles.derive: a row or an alias collapses to the row; any other name keeps its own advertised spelling)."""
models = (accounts or {}).get('models') or {}
aliases = (accounts or {}).get('aliases') or {}
if name in models:
return name
return aliases[name] if aliases.get(name) in models else name
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |