boomkin 0.9.0 → 0.10.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,20 @@
1
1
  # Changelog
2
2
 
3
+ ## 0.10.0 · 2026-10-09
4
+
5
+ - Add four reviewed project templates with explicit defaults and required inputs, frozen per-project context and previews without model or data calls.
6
+ - Add public Hyperliquid wallet audits, market research, completed-trade review and frozen strategy validation workflows, expanding discovery to 31 tasks.
7
+ - Add research rule presets and separate held-out/higher-cost validation to the browser strategy lab, with downloadable reports and disclosed restart accounting.
8
+ - Review independently released Hyperliquid 0.4.0 and strategy 0.2.0 resources; preserve immutable source pins and existing profile selections.
9
+
10
+ ## 0.9.1 · 2026-10-08
11
+
12
+ - Refresh the GitHub and npm introduction with a browser demo, clear installation steps and a public-data first task.
13
+ - Add human and agent documentation indexes, accessible README artwork and current discovery metadata.
14
+ - Update the reviewed catalog to independently published documentation patches for all 17 skill packs.
15
+ - Route Hyperliquid account inspection to its account-monitoring skill.
16
+ - Keep npm commands and the website setup guide aligned; fix the strategy lab heading after an invalid run.
17
+
3
18
  ## 0.9.0 · 2026-10-08
4
19
 
5
20
  - Add local discovery of 60 providers with primary-source provenance, capability, access, cost and transaction authority metadata.
package/README.md CHANGED
@@ -2,210 +2,152 @@
2
2
 
3
3
  **Your open-source DeFi desk.**
4
4
 
5
- Inspect an Aave position, track a Lido withdrawal, size a Uniswap range, check Aerodrome gauge custody or screen DefiLlama yields. Boomkin installs protocol-specific procedures and local helpers into the native [Hermes Agent](https://github.com/NousResearch/hermes-agent), with reviewed source pins and file-integrity checks.
5
+ Research markets, inspect positions and test a thesis with a native [Hermes agent](https://github.com/NousResearch/hermes-agent). Keep the sources, assumptions and run history in your own profile.
6
6
 
7
- Each workflow names the reads to make, the accounting that matters and the evidence needed to finish. Official provider tools do the protocol work; Hermes handles the agent loop.
7
+ [![Boomkin: research, inspect and test your DeFi thesis](https://raw.githubusercontent.com/galleonlabs/boomkin/main/docs/assets/boomkin-hero.svg)](https://galleonlabs.github.io/boomkin/lab/)
8
8
 
9
- [Website](https://galleonlabs.github.io/boomkin/) · [Strategy lab](https://galleonlabs.github.io/boomkin/lab/) · [Documentation](https://galleonlabs.github.io/boomkin/docs/) · [Contribute](CONTRIBUTING.md)
9
+ **[Try the strategy lab →](https://galleonlabs.github.io/boomkin/lab/)** · [Explore workflows](https://galleonlabs.github.io/boomkin/#workflows) · [Read the docs](https://galleonlabs.github.io/boomkin/docs/)
10
+
11
+ The browser demo needs no installation. Choose a research rule, compare held-out periods and higher trading costs, and inspect simulated trades using a dated Bitcoin capture, synthetic prices or your own dataset. Uploaded data stays in your browser.
12
+
13
+ **31 workflows · 60 providers in discovery · 17 independent packs · 49 reviewed skills**
10
14
 
11
15
  [![CI](https://github.com/galleonlabs/boomkin/actions/workflows/ci.yml/badge.svg)](https://github.com/galleonlabs/boomkin/actions/workflows/ci.yml)
16
+ [![npm](https://img.shields.io/npm/v/boomkin)](https://www.npmjs.com/package/boomkin)
12
17
  [![MIT](https://img.shields.io/badge/license-MIT-blue)](LICENSE)
13
18
 
14
19
  ## Get started
15
20
 
16
- Use macOS or Linux, [Bun](https://bun.sh), Git and a terminal. For Windows, use WSL2 or install Hermes with its official Windows flow first. Optional Coinbase tooling requires Node.js 22+; Agentic Wallet's current guide requires Node.js 24+.
21
+ Use macOS or Linux with [Bun](https://bun.sh) 1.3.12+ and Git. Windows users can use WSL2.
17
22
 
18
23
  ```bash
19
- bun install -g boomkin@0.9.0
24
+ bun install -g boomkin@0.10.0
20
25
  boomkin onboard
21
26
  boomkin doctor --live
22
27
  boomkin start
23
28
  ```
24
29
 
25
- Bun and Git are prerequisites. Keep Bun's global bin directory on your path. For a source checkout:
26
-
27
- ```bash
28
- git clone https://github.com/galleonlabs/boomkin.git
29
- cd boomkin
30
- bun install --frozen-lockfile
31
- bun run boomkin onboard
32
- bun run boomkin doctor --live
33
- bun run boomkin start
34
- ```
35
-
36
- `onboard` prepares `~/.boomkin/hermes`, installs the reviewed official Hermes runtime if none is available, installs all packs in the reviewed catalog on a fresh profile, creates Boomkin's SOUL and instructions, configures public CoinGecko MCP, and opens native Hermes model setup. Choose your provider and sign in there. `start` launches the native Hermes chat in that same profile.
30
+ Onboarding prepares `~/.boomkin/hermes`, installs the reviewed packs, configures public CoinGecko data and opens native Hermes model setup. Choose your model provider and sign in there. `start` opens chat in that profile.
37
31
 
38
- Hermes remains the runtime: its tools, sessions, memory, model adapters, MCP support and scheduling are not forked. A new runtime installation uses reviewed Hermes v0.21.5 source (upstream tag `v2026.9.24`); the native installer also manages the user-level `hermes` command and dependencies. Browser/computer-use dependencies are skipped initially and can be added through native Hermes setup. An existing working Hermes installation at 0.21.0 or newer is reused, not downgraded.
32
+ CoinGecko is the sole default connection. Model access uses your account or local configuration; optional providers can require credentials, subscriptions or paid requests. Setup creates no wallet, makes no model call and starts no background service.
39
33
 
40
- Model authentication requires your account or local model configuration. Optional providers can require subscriptions, OAuth or scoped credentials. Onboarding makes no model call, pays for no data, creates or funds no wallet, and starts no background service.
34
+ [Profiles, install options and source setup](docs/GETTING-STARTED.md) · [Model connections](docs/CONNECTIONS.md)
41
35
 
42
- Useful options:
36
+ ## First useful task
43
37
 
44
- ```bash
45
- bun run boomkin onboard --directory "$HOME/defi-agent" --dry-run
46
- bun run boomkin onboard --directory "$HOME/defi-agent" --skip-model-setup
47
- bun run boomkin onboard --directory "$HOME/defi-agent" --no-install
48
- bun run boomkin model --directory "$HOME/defi-agent"
49
- ```
38
+ Paste this into your new Hermes chat:
50
39
 
51
- Use the same `--directory` for subsequent commands. `--skip-model-setup` is for preparing a profile before interactive login; it does not mark the model authenticated. Existing SOUL, instructions and unrelated settings are preserved. A conflicting named MCP configuration is reported for review instead of overwritten.
40
+ > Use galleon-coingecko-token-research. Resolve CoinGecko ID ethereum and capture a public ETH/USD market observation. Show the asset identity, source URL, observation time and freshness limits. Mark missing data clearly and keep this task read-only.
52
41
 
53
- ## Keep the question and the evidence
42
+ The result should separate supported observations from unknowns. `doctor --live` establishes public tool discovery; this first task checks whether your configured model can retrieve and explain market evidence.
54
43
 
55
- Save a research question as a project, preview its brief and run it through native Hermes. Each run keeps its own report, raw sources, observation hashes and reviewable unsigned plan. A model response stays in review until its artifact checks pass. [Project commands and output contract](docs/PROJECTS.md).
44
+ ## Start from a research template
56
45
 
57
46
  ```bash
58
- boomkin project create --name stablecoin-yields \
59
- --workflow yield-screen --input-file ./inputs.json
60
- boomkin project run --name stablecoin-yields --dry-run
61
- boomkin project run --name stablecoin-yields
62
- boomkin project check --name stablecoin-yields
47
+ boomkin templates
48
+ boomkin templates --template wallet-audit --json
49
+ boomkin project create --name wallet-review \
50
+ --template wallet-audit --input-file ./inputs.json
51
+ boomkin project run --name wallet-review --dry-run
63
52
  ```
64
53
 
65
- Public market snapshots compare fresh CoinGecko and DefiLlama observations while keeping unavailable, stale and disagreeing sources visible. The strategy pack runs local daily spot simulations with next-observation decisions, fixed fees and adverse slippage, flow-neutral returns, drawdown and a same-flow buy-and-hold benchmark. [Try the same engine in the browser](https://galleonlabs.github.io/boomkin/lab/) with a dated Bitcoin capture, synthetic fixtures or your own dataset.
54
+ Four reviewed templates cover public wallet cost audits, frozen strategy validation, stablecoin income and thesis review. Supply your public account/window or local dataset/rule paths; required fields and defaults are visible before saving. Templates preserve the question and evidence in your own profile. [Inputs and run guide](docs/PROJECTS.md)
66
55
 
67
56
  ## Pick a job
68
57
 
69
58
  ```bash
70
- bun run boomkin workflows
71
- bun run boomkin workflows --protocol Uniswap
72
- bun run boomkin workflows --workflow aave-health --json
73
- bun run boomkin onboard --workflow aave-health
59
+ boomkin workflows
60
+ boomkin workflows --protocol Aave
61
+ boomkin workflows --workflow transaction-simulation --json
62
+ boomkin onboard --workflow aave-health
74
63
  ```
75
64
 
76
- `workflows` works before setup and makes no network calls. It shows the required inputs, deliverable and provider access. `--workflow` installs that task's independent pack. On an existing profile it selects that pack for future updates, just like `--pack`; already copied files and custom instructions are preserved. Use a separate `--directory` for a dedicated task profile.
65
+ Workflow discovery runs locally. Each job names its inputs, expected result and provider requirements. Selecting a workflow installs its pack and saves that selection for updates. Already copied skills remain; choose a separate `--directory` for a dedicated profile.
77
66
 
78
- | Task | What you get |
67
+ | You want to… | Start with |
79
68
  | --- | --- |
80
- | `market-snapshot` | Source timestamps, raw-response hashes and public price comparison |
81
- | `strategy-backtest` | Reproducible daily spot tests, costs and a same-flow benchmark |
82
- | `thesis-review` | Saved thesis, invalidation conditions and fresh evidence for review |
83
- | `aave-health` | V3 position reads, stressed health factor and reserve constraints |
84
- | `morpho-market` / `compound-borrow` | Protocol-specific collateral, debt and liquidity analysis |
85
- | `uniswap-position` / `aerodrome-position` | Position accounting, range checks and gauge/exit requirements |
86
- | `uniswap-quote` / `bridge-status` | Route review, permissions and settlement evidence |
87
- | `lido-exit` | Withdrawal NFT ownership, finalization and claimability |
88
- | `pendle-maturity` / `vault-exit` | Maturity accounting and what can actually be redeemed |
89
- | `token-research` / `yield-screen` | Contract-based identity and reproducible yield screening |
90
- | `coinbase-readiness` | CDP/AgentKit capabilities and policy gaps |
91
- | `hyperliquid-risk` | Venue/account-mode, margin and funding analysis |
92
- | `agent-capabilities` | Exact access, cost and permission boundaries for a tool |
93
- | `unsigned-plan` / `transaction-simulation` | Ordered signer-free payload construction and exact-state simulation evidence |
94
- | `aave-v4-health` | Verified V4 hub/spoke accounting and deployment maturity |
95
- | `sablier-stream` / `superfluid-stream` | Vesting rights, claimable amounts, flow deposits and runway |
96
- | `across-route` / `explorer-diligence` | Cross-chain quote/arrival and code/control/transaction evidence |
97
- | `prediction-research` / `prediction-resolution` | Public Polymarket research, resolution and payout constraints |
98
-
99
- ## Discover the right provider
69
+ | Capture market evidence or test a rule | `market-snapshot`, `strategy-backtest` |
70
+ | Inspect lending or an exit | `aave-health`, `aave-v4-health`, `morpho-market`, `lido-exit` |
71
+ | Review liquidity and routes | `uniswap-position`, `aerodrome-position`, `across-route` |
72
+ | Screen income or streaming obligations | `yield-screen`, `sablier-stream`, `superfluid-stream` |
73
+ | Review an exact unsigned payload | `unsigned-plan`, `transaction-simulation` |
74
+ | Research prediction markets | `prediction-research`, `prediction-resolution` |
100
75
 
101
- ```bash
102
- boomkin providers --search simulation --json
103
- boomkin providers --capability Data --access keyless
104
- boomkin providers --provider blockscout
105
- boomkin connect --provider blockscout
106
- ```
107
-
108
- Discovery covers all 58 studied [skills.eth.sh](https://skills.eth.sh/) entries and two existing connections with dated primary sources, authorship, access, cost and authority metadata. It runs locally and installs no third-party skills. Keyless access and free access are distinct; wallet tools and paid x402 services remain explicit choices. CoinGecko remains the sole onboarding default; Blockscout is an optional keyless explorer with reviewed named reads. [Provider guide](docs/PROVIDERS.md).
109
-
110
- ## Skill packs
76
+ [All 31 workflows, inputs and outputs](docs/WORKFLOWS.md)
111
77
 
112
- The reviewed catalog contains **17 independent packs and 48 skills**. Install all of them or choose the workflows you need; the agent loads relevant instructions on demand.
113
-
114
- | Pack | What it covers |
115
- | --- | --- |
116
- | [Agent plans](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/agent) | Unsigned builder/playbook plans and exact simulation evidence |
117
- | [Prediction markets](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/prediction) | Rules, order-book depth, versioned ledgers and payout evidence |
118
- | [Infrastructure](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/infra) | RPC, Alchemy, Coinbase, wallet policies and readiness |
119
- | [Data](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/data) | CoinGecko, DefiLlama and AIXBT research, identity, freshness and source-hashed market captures |
120
- | [Strategy](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/strategy) | Daily spot simulations, cash flows, costs, drawdown and reproducible benchmarks |
121
- | [Liquidity provision](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/lp) | Uniswap, Aerodrome, Revert and VFAT position workflows |
122
- | [Hyperliquid](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/hyperliquid) | Venue-specific analysis, planning, execution, monitoring and review |
123
- | [Lending](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/lending) | Collateral, borrowing, interest, repayment and liquidation risk |
124
- | [Staking](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/staking) | Staking, restaking, receipts and withdrawal queues |
125
- | [Yield](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/yield) | Vault diligence, share accounting and yield sources |
126
- | [Tokenized assets](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/tokenized-assets) | Asset claims, eligibility, restrictions and redemption |
127
- | [Routing](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/routing) | Swap and bridge quotes, route limits and settlement |
128
- | [Derivatives](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/derivatives) | Exposure, margin, funding and venue-specific constraints |
129
- | [Portfolio](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/portfolio) | Assets, debt, cash flows, performance and rebalance plans |
130
- | [Security](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/security) | Token diligence, evidence comparisons, transaction effects and permissions |
131
- | [Payments](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/payments) | Agent payments, x402 challenges and settlement evidence |
132
- | [Governance](https://github.com/galleonlabs/crypto-defi-skills/tree/main/packages/governance) | Proposals, voting, delegation and execution stages |
133
-
134
- Start with the protocol workflow matching your task. Use infrastructure and cross-protocol skills when the task needs them. The broader primitive skills use the `galleon-defi-` prefix to avoid upstream name collisions. Read [workflow examples](docs/WORKFLOWS.md) for useful combinations.
135
-
136
- Each pack is independently versioned and published from [crypto-defi-skills](https://github.com/galleonlabs/crypto-defi-skills). Boomkin's [catalog](catalog/skills.json) records the npm identity, version, immutable source commit, package directory and expected skills. Downloads, installed metadata and every copied supporting file are checked before a successful sync is recorded. Local file digests let `check` and `doctor` detect missing references and changed scripts afterward.
137
-
138
- Fresh onboarding includes every pack in the checked-out catalog. Select fewer with repeated `--pack` options:
78
+ ## Keep the question and the evidence
139
79
 
140
- ```bash
141
- bun run boomkin onboard --pack defi-infra-skills --pack defi-data-skills
80
+ Save this public research example as `inputs.json`:
81
+
82
+ ```json
83
+ {
84
+ "chain": "Ethereum",
85
+ "asset": "USDC",
86
+ "minimumTvlUsd": "10000000",
87
+ "yieldPreference": "Separate base yield from incentive emissions",
88
+ "limit": 5,
89
+ "question": "What evidence is missing before a deposit decision?"
90
+ }
142
91
  ```
143
92
 
144
- Updates preserve your selection. Future packs are opt-in. To expand an existing installation deliberately:
145
-
146
93
  ```bash
147
- bun run boomkin onboard --directory "$HOME/your-existing-hermes-profile" --all-packs
94
+ boomkin project create --name stablecoin-yields \
95
+ --workflow yield-screen --input-file ./inputs.json
96
+ boomkin project run --name stablecoin-yields --dry-run
97
+ boomkin onboard --workflow yield-screen --skip-model-setup
98
+ boomkin project run --name stablecoin-yields
99
+ boomkin project check --name stablecoin-yields
148
100
  ```
149
101
 
150
- Other harnesses can still install the portable skills through [advanced compatibility setup](docs/HARNESSES.md). Boomkin's end-to-end product flow is Hermes-first.
102
+ Creation and preview make no model or data calls. Onboarding prepares the selected workflow and preserves your configured model. A live run uses that model and your tools, then saves a separate report, raw sources, hashes and unsigned plan. Checks establish the saved evidence contract; they do not certify economic safety or authorize execution. [Project guide](docs/PROJECTS.md)
151
103
 
152
- ## Connect your tools
104
+ ## Discover the right provider
153
105
 
154
106
  ```bash
155
- bun run boomkin providers
156
- bun run boomkin connect --provider aixbt
157
- bun run boomkin connect --provider alchemy
158
- bun run boomkin connect --provider defillama
159
- bun run boomkin connect --provider coinbase
160
- bun run boomkin connect --provider tenderly
107
+ boomkin providers --search simulation --json
108
+ boomkin providers --capability Data --access keyless
109
+ boomkin providers --provider blockscout
161
110
  ```
162
111
 
163
- | Connection | Setup and scope |
164
- | --- | --- |
165
- | CoinGecko | Public, keyless MCP configured during onboarding. Its reviewed tools are `execute` and `search_docs`; hosted execution is restricted to the provider's data SDK. |
166
- | Alchemy | Native Hermes OAuth and explicit tool selection. Select the intended Alchemy app; data/RPC access and wallet/admin actions have different scopes. |
167
- | AIXBT | Optional crypto intelligence through native Hermes MCP; the key stays in `AIXBT_API_KEY`. Discovery is public; protected research reads require account access. |
168
- | DeFiLlama | Native Hermes OAuth with an API subscription. Queries consume credits; connecting another client can disconnect the previous client. |
169
- | Tenderly | Optional paid-plan OAuth with explicit tool selection. Simulations and inspection require the intended project; results persist there. |
170
- | Coinbase account | Official local MCP through a pinned CLI, with six read tools and a profile-specific configuration/keychain environment. Account credentials remain a separate step. |
171
- | Agentic Wallet | Separate official `awal` CLI flow for managed wallet and x402 use. Login, wallet creation, funding and spend limits require your choices; see the connection guide. |
112
+ The directory covers 58 studied [skills.eth.sh](https://skills.eth.sh/) entries and two existing connections. Dated sources distinguish authorship, access, cost and authority. Discovery does not install or connect all 60 providers. Keyless access can still cost money; wallet and x402 capabilities need separate choices. [Provider guide](docs/PROVIDERS.md)
172
113
 
173
- For Coinbase, supply a scoped key file to the native CLI through Boomkin:
114
+ ## Connect your tools
174
115
 
175
116
  ```bash
176
- bun run boomkin connect --provider coinbase --key-file /absolute/path/to/scoped-key.json
117
+ boomkin connect --provider blockscout
118
+ boomkin doctor --live
177
119
  ```
178
120
 
179
- The command uses the OS keychain and does not opt into plaintext secret storage. It configures credentials; it does not check balances, trade or approve payments. Coinbase's remote account MCP currently restricts custom harnesses, so Boomkin uses its supported local server. Coinbase for Agents and Agentic Wallet have different custody and payment capabilities.
121
+ Blockscout adds optional keyless explorer reads. Other supported connections include AIXBT, Alchemy, DeFiLlama, Tenderly and Coinbase account data, each with its own setup and access boundary. Restart Hermes after changing MCP settings.
180
122
 
181
- Read [the connection guide](docs/CONNECTIONS.md) for exact prerequisites, diagnostics and limits. Restart Hermes after changing MCP settings. MCP tool selection controls that connection; it is not a sandbox around Hermes's terminal or a substitute for provider-enforced account and spending limits.
123
+ Hermes owns tools and authentication. MCP tool selection does not constrain its terminal or replace provider-enforced account and spending limits. [Exact prerequisites and connection scope](docs/CONNECTIONS.md)
182
124
 
183
- ## First useful task
125
+ ## Skill packs
184
126
 
185
- > Use galleon-aave-position to inspect my wallet on this Aave V3 market: [chain, market and address]. Show current health factor and the effect of a 20% collateral-price fall. Identify missing reads and keep any repayment or borrow plan unsigned.
127
+ Boomkin installs reviewed releases from [crypto-defi-skills](https://github.com/galleonlabs/crypto-defi-skills): agent plans, prediction markets, infrastructure, data, strategy, liquidity, Hyperliquid, lending, staking, yield, tokenized assets, routing, derivatives, portfolio, security, payments and governance.
186
128
 
187
- Boomkin loads only the relevant skills and provider references. It separates infrastructure readiness, data evidence, protocol workflows, portfolio reporting and transaction review from execution. It uses Hermes's native memory and scheduling when appropriate; a scheduled task retains the same authorization and freshness requirements as an interactive task.
129
+ Each pack installs independently. The [catalog](catalog/skills.json) pins its package, version, immutable source commit and expected skills. Installation checks the copied supporting files; later integrity checks reveal missing or modified instructions and scripts.
188
130
 
189
- `doctor` reports configuration, installed versions and file integrity. Older installations can run `update` once to record file digests; preserve any local skill edits first. `doctor --live` additionally verifies keyless CoinGecko MCP initialization and tool discovery, plus credential-free AIXBT discovery when connected. Neither is proof of a successful model response, authenticated paid data, wallet authority or an executed transaction. Use the data pack's public diagnostic for a first market observation, and the infrastructure pack's RPC diagnostic for a configured chain.
131
+ Fresh onboarding includes all 17 packs. Repeated `--pack` options choose fewer; updates preserve that selection and future packs remain opt-in. Hermes supplies the native runtime, sessions, memory and model setup. [Other harnesses](docs/HARNESSES.md) receive portable skills only.
190
132
 
191
133
  ## Updates and recovery
192
134
 
193
135
  ```bash
194
- bun run boomkin check --directory "$HOME/.boomkin/hermes"
195
- bun run boomkin update --directory "$HOME/.boomkin/hermes"
196
- bun run boomkin doctor --live
197
- bun run boomkin start
136
+ bun install -g boomkin@0.10.0
137
+ boomkin check --directory "$HOME/.boomkin/hermes"
138
+ boomkin update --directory "$HOME/.boomkin/hermes"
139
+ boomkin doctor --directory "$HOME/.boomkin/hermes" --live
198
140
  ```
199
141
 
200
- Pull this repo and run `bun install --frozen-lockfile` when updating Boomkin itself. `update` refreshes the reviewed skill catalog, not the Hermes runtime. Use native Hermes updates separately and rerun the readiness checks afterward. The reviewed runtime pin and installer checksum are in [src/hermes.ts](src/hermes.ts).
142
+ Update Boomkin, selected skill packs and Hermes separately. Preserve local skill edits before updates. [Updates, backups and recovery](docs/UPDATES.md)
201
143
 
202
- Onboarding can be rerun after a failed install or interrupted login. Completed skill installs and existing instructions remain intact. Review any reported operation lock before removing it. Provider failure does not change the approved route or cause an automatic retry of a financial action.
144
+ ## ChatGPT authentication
203
145
 
204
- See [updates and recovery](docs/UPDATES.md) for pack selection, backups and troubleshooting. Keep credentials, private evidence and generated profiles outside this repository.
146
+ Use `boomkin chatgpt --action login` and `--action status` for native Hermes ChatGPT OAuth. Model selection stays with Hermes. [The authentication guide](docs/CHATGPT.md) distinguishes sign-in, partner plan consent and verified usage attribution.
205
147
 
206
148
  ## Contributing
207
149
 
208
- Bug reports, clearer guides, new workflows and focused pull requests are welcome. Start with [CONTRIBUTING.md](CONTRIBUTING.md) for setup and the checks relevant to your change. Report a reproducible problem in [GitHub Issues](https://github.com/galleonlabs/boomkin/issues); use [private reporting](SECURITY.md) for security concerns.
150
+ Bug reports, clearer guides and focused pull requests are welcome. [Contributing guide](CONTRIBUTING.md) · [Issues](https://github.com/galleonlabs/boomkin/issues) · [Private security reporting](SECURITY.md)
209
151
 
210
152
  ### Local development
211
153
 
@@ -213,19 +155,12 @@ Bug reports, clearer guides, new workflows and focused pull requests are welcome
213
155
  bun install --frozen-lockfile
214
156
  bun run check
215
157
  bun run build
216
- bun run smoke
217
158
  ```
218
159
 
219
- The compatibility smoke checks fresh setup/update and independent pack selection. [Native Hermes smoke](scripts/hermes-native-smoke.py) additionally checks the actual runtime's profile, SOUL, MCP configuration and tool filtering in a temporary home, using a local mock MCP. `--public` adds a keyless CoinGecko discovery check. No model or wallet action is part of these checks.
160
+ Run the [change-specific checks](CONTRIBUTING.md#work-locally) before a pull request. Agents can start with [the documentation index](docs/AGENT-INDEX.md) and [AGENTS.md](AGENTS.md).
220
161
 
221
162
  ## License and credit
222
163
 
223
- [MIT licensed](LICENSE), with the copyright and permission notice retained when reusing copies or substantial portions. Created by [Andrew Wilkinson](https://andrewwilkinson.io) and [Galleon Labs](https://github.com/galleonlabs).
224
-
225
- See [reuse and attribution](ATTRIBUTION.md) for a ready-to-copy credit line. If Boomkin helps your work, [a star on the original repository](https://github.com/galleonlabs/boomkin) is appreciated and entirely optional.
226
-
227
- The documentation copy controls were inspired by [Rare UI](https://www.rareui.com/components/code-block) and implemented independently.
228
-
229
- ## ChatGPT authentication
164
+ [MIT licensed](LICENSE). Created by [Andrew Wilkinson](https://andrewwilkinson.io) and [Galleon Labs](https://github.com/galleonlabs). Preserve copyright and license notices when reusing the project; [attribution guidance](ATTRIBUTION.md) includes an optional credit line. A repository star is appreciated and entirely optional.
230
165
 
231
- Use `bun run boomkin chatgpt --action login` for native Hermes ChatGPT OAuth, `--action status` to inspect it, and `--action model` for catalog-supported selection including GPT-6.1 Sol when available. [ChatGPT onboarding and verification](docs/CHATGPT.md) distinguishes native OAuth from partner plan consent and attributed usage. No tokens are copied into Boomkin and the runtime pin is unchanged.
166
+ Documentation copy controls were inspired by [Rare UI](https://www.rareui.com/components/code-block) and implemented independently.
@@ -5,8 +5,8 @@
5
5
  "id": "lp-skills",
6
6
  "source": "galleonlabs/crypto-defi-skills",
7
7
  "description": "Agent skills and deterministic tools for Uniswap and Aerodrome liquidity provision.",
8
- "version": "0.6.0",
9
- "revision": "8609564d3a79883ebf515fec201240e90fb271b2",
8
+ "version": "0.6.1",
9
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
10
10
  "skills": [
11
11
  "aerodrome-slipstream",
12
12
  "lp-analyze",
@@ -23,9 +23,9 @@
23
23
  {
24
24
  "id": "hyperliquid-skills",
25
25
  "source": "galleonlabs/crypto-defi-skills",
26
- "description": "Agent skills for Hyperliquid setup, market analysis, trade planning and execution, monitoring, and performance review.",
27
- "version": "0.3.4",
28
- "revision": "b4a020cbc810acabe22a09662cf88137a0907ea6",
26
+ "description": "Hyperliquid market research, trade planning, monitoring, performance review and reproducible public wallet cost audits.",
27
+ "version": "0.4.0",
28
+ "revision": "3f91a955d59a55b9927403ddc3c2e26660d48bf8",
29
29
  "skills": [
30
30
  "hyperliquid-analyze",
31
31
  "hyperliquid-engineer",
@@ -33,7 +33,8 @@
33
33
  "hyperliquid-monitor",
34
34
  "hyperliquid-plan",
35
35
  "hyperliquid-review",
36
- "hyperliquid-setup"
36
+ "hyperliquid-setup",
37
+ "hyperliquid-wallet-audit"
37
38
  ],
38
39
  "path": "packages/hyperliquid",
39
40
  "package": "galleon-hyperliquid-skills"
@@ -42,8 +43,8 @@
42
43
  "id": "defi-infra-skills",
43
44
  "source": "galleonlabs/crypto-defi-skills",
44
45
  "description": "RPC, wallet and Hermes infrastructure skills with read-only readiness diagnostics.",
45
- "version": "0.4.0",
46
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
46
+ "version": "0.4.1",
47
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
47
48
  "skills": [
48
49
  "galleon-coinbase-agentkit-readiness",
49
50
  "galleon-defi-infra"
@@ -55,8 +56,8 @@
55
56
  "id": "defi-data-skills",
56
57
  "source": "galleonlabs/crypto-defi-skills",
57
58
  "description": "Capture timestamped market evidence, compare public sources, retrieve daily history and screen DeFi yields.",
58
- "version": "0.6.0",
59
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
59
+ "version": "0.6.1",
60
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
60
61
  "skills": [
61
62
  "galleon-coingecko-token-research",
62
63
  "galleon-defi-data",
@@ -70,8 +71,8 @@
70
71
  "id": "defi-lending-skills",
71
72
  "source": "galleonlabs/crypto-defi-skills",
72
73
  "description": "Research lending markets, plan borrowing and monitor liquidation exposure.",
73
- "version": "0.3.0",
74
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
74
+ "version": "0.3.1",
75
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
75
76
  "skills": [
76
77
  "galleon-aave-position",
77
78
  "galleon-aave-v4",
@@ -86,8 +87,8 @@
86
87
  "id": "defi-staking-skills",
87
88
  "source": "galleonlabs/crypto-defi-skills",
88
89
  "description": "Plan liquid staking, restaking and queued withdrawals.",
89
- "version": "0.2.0",
90
- "revision": "8609564d3a79883ebf515fec201240e90fb271b2",
90
+ "version": "0.2.1",
91
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
91
92
  "skills": [
92
93
  "galleon-defi-staking",
93
94
  "galleon-lido-withdrawals"
@@ -99,8 +100,8 @@
99
100
  "id": "defi-yield-skills",
100
101
  "source": "galleonlabs/crypto-defi-skills",
101
102
  "description": "Evaluate vaults and yield tokens, including liquidity and maturity.",
102
- "version": "0.2.0",
103
- "revision": "8609564d3a79883ebf515fec201240e90fb271b2",
103
+ "version": "0.2.1",
104
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
104
105
  "skills": [
105
106
  "galleon-defi-yield",
106
107
  "galleon-pendle-maturity",
@@ -113,8 +114,8 @@
113
114
  "id": "defi-tokenized-assets-skills",
114
115
  "source": "galleonlabs/crypto-defi-skills",
115
116
  "description": "Research tokenized assets, eligibility and issuer settlement.",
116
- "version": "0.1.2",
117
- "revision": "d5d8439bd5bd1fec88994a6ca0448cde11e31747",
117
+ "version": "0.1.3",
118
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
118
119
  "skills": [
119
120
  "galleon-defi-tokenized-assets"
120
121
  ],
@@ -125,8 +126,8 @@
125
126
  "id": "defi-routing-skills",
126
127
  "source": "galleonlabs/crypto-defi-skills",
127
128
  "description": "Compare swaps and bridges and track cross-chain settlement.",
128
- "version": "0.3.0",
129
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
129
+ "version": "0.3.1",
130
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
130
131
  "skills": [
131
132
  "galleon-defi-routing",
132
133
  "lifi-cross-chain",
@@ -139,8 +140,8 @@
139
140
  "id": "defi-derivatives-skills",
140
141
  "source": "galleonlabs/crypto-defi-skills",
141
142
  "description": "Plan derivatives positions and reconcile venue execution.",
142
- "version": "0.1.4",
143
- "revision": "470508005adb1ed3fe3a1e758598170e671b7db3",
143
+ "version": "0.1.5",
144
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
144
145
  "skills": [
145
146
  "galleon-defi-derivatives"
146
147
  ],
@@ -151,8 +152,8 @@
151
152
  "id": "defi-portfolio-skills",
152
153
  "source": "galleonlabs/crypto-defi-skills",
153
154
  "description": "Reconcile DeFi assets, liabilities, cash flows and exposure.",
154
- "version": "0.2.0",
155
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
155
+ "version": "0.2.1",
156
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
156
157
  "skills": [
157
158
  "galleon-defi-portfolio"
158
159
  ],
@@ -163,8 +164,8 @@
163
164
  "id": "defi-security-skills",
164
165
  "source": "galleonlabs/crypto-defi-skills",
165
166
  "description": "Investigate token risks, compare evidence and review transaction permissions.",
166
- "version": "0.2.2",
167
- "revision": "d5d8439bd5bd1fec88994a6ca0448cde11e31747",
167
+ "version": "0.2.3",
168
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
168
169
  "skills": [
169
170
  "galleon-defi-security",
170
171
  "galleon-defi-security-token-diligence"
@@ -176,8 +177,8 @@
176
177
  "id": "defi-payments-skills",
177
178
  "source": "galleonlabs/crypto-defi-skills",
178
179
  "description": "Plan stablecoin payments, x402 purchases and token streams.",
179
- "version": "0.2.0",
180
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
180
+ "version": "0.2.1",
181
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
181
182
  "skills": [
182
183
  "galleon-defi-payments",
183
184
  "galleon-sablier-streams",
@@ -190,8 +191,8 @@
190
191
  "id": "defi-governance-skills",
191
192
  "source": "galleonlabs/crypto-defi-skills",
192
193
  "description": "Research proposals, prepare votes and track governance execution.",
193
- "version": "0.1.2",
194
- "revision": "d5d8439bd5bd1fec88994a6ca0448cde11e31747",
194
+ "version": "0.1.3",
195
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
195
196
  "skills": [
196
197
  "galleon-defi-governance"
197
198
  ],
@@ -201,9 +202,9 @@
201
202
  {
202
203
  "id": "defi-strategy-skills",
203
204
  "source": "galleonlabs/crypto-defi-skills",
204
- "description": "Test daily spot rules with costs, cash flows, drawdown and a reproducible buy-and-hold benchmark.",
205
- "version": "0.1.0",
206
- "revision": "72b0d85df41c2c2a6e0525d94b245faa6d11e5d4",
205
+ "description": "Test frozen daily spot rules with chronological held-out periods, higher trading costs and reproducible same-cost benchmarks.",
206
+ "version": "0.2.0",
207
+ "revision": "3f91a955d59a55b9927403ddc3c2e26660d48bf8",
207
208
  "skills": [
208
209
  "galleon-defi-strategy-backtest"
209
210
  ],
@@ -214,8 +215,8 @@
214
215
  "id": "defi-agent-skills",
215
216
  "source": "galleonlabs/crypto-defi-skills",
216
217
  "description": "Construct unsigned DeFi plans and verify exact transaction simulation evidence.",
217
- "version": "0.1.0",
218
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
218
+ "version": "0.1.1",
219
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
219
220
  "skills": [
220
221
  "galleon-defi-agent-plan",
221
222
  "galleon-defi-agent-simulate"
@@ -227,8 +228,8 @@
227
228
  "id": "defi-prediction-skills",
228
229
  "source": "galleonlabs/crypto-defi-skills",
229
230
  "description": "Research prediction market order books and reconcile resolution without signing or trading.",
230
- "version": "0.1.0",
231
- "revision": "7451b2cb578ce786cbd1e3ee55a4d12ba599739d",
231
+ "version": "0.1.1",
232
+ "revision": "9fe67c24efc0fd5796066605b29576f3b8f2de94",
232
233
  "skills": [
233
234
  "galleon-prediction-market-research",
234
235
  "galleon-prediction-market-resolution"
@@ -0,0 +1,41 @@
1
+ {
2
+ "schemaVersion": 1,
3
+ "templates": [
4
+ {
5
+ "id": "wallet-audit",
6
+ "title": "Find the costs in a Hyperliquid wallet",
7
+ "workflow": "hyperliquid-wallet-audit",
8
+ "description": "Reconcile a public account's realized activity, fees, funding and current exposure. Preserve raw sources and explain incomplete history.",
9
+ "defaults": { "network": "mainnet", "maxPages": 10 },
10
+ "requiredInputs": ["account", "startTime", "endTime"],
11
+ "exampleInputs": { "account": "YOUR_PUBLIC_ACCOUNT_ADDRESS", "startTime": "2026-10-01T00:00:00.000Z", "endTime": "2026-10-08T00:00:00.000Z" }
12
+ },
13
+ {
14
+ "id": "strategy-validation",
15
+ "title": "Test a rule on an untouched period",
16
+ "workflow": "strategy-validation",
17
+ "description": "Freeze a daily spot rule, compare higher trading costs and inspect a separate held-out period against the same-cost benchmark.",
18
+ "defaults": {},
19
+ "requiredInputs": ["dataFile", "specFile"],
20
+ "exampleInputs": { "dataFile": "/absolute/path/to/daily-prices.json", "specFile": "/absolute/path/to/frozen-rule.json" }
21
+ },
22
+ {
23
+ "id": "stablecoin-yields",
24
+ "title": "Screen stablecoin income",
25
+ "workflow": "yield-screen",
26
+ "description": "Separate base income from incentives and collect the missing exit and liquidity evidence before a deposit decision.",
27
+ "defaults": { "minimumTvlUsd": "10000000", "limit": 5, "yieldPreference": "Separate base yield from incentive emissions" },
28
+ "requiredInputs": ["chain", "asset"],
29
+ "exampleInputs": { "chain": "Ethereum", "asset": "USDC" }
30
+ },
31
+ {
32
+ "id": "thesis-review",
33
+ "title": "Keep an investment thesis accountable",
34
+ "workflow": "thesis-review",
35
+ "description": "Save explicit invalidation conditions and review preferences, then compare each research run with its captured baseline.",
36
+ "defaults": { "reviewFrequency": "Weekly", "notifyWhen": "An invalidation condition changes verdict" },
37
+ "requiredInputs": ["asset", "thesis", "invalidationConditions", "evidenceStandard"],
38
+ "exampleInputs": { "asset": "YOUR_EXACT_ASSET_IDENTITY", "thesis": "YOUR_THESIS", "invalidationConditions": ["YOUR_MEASURABLE_INVALIDATION_CONDITION"], "evidenceStandard": "YOUR_PRIMARY_SOURCE_REQUIREMENTS" }
39
+ }
40
+ ]
41
+ }