olier_batch_search_plants
ChatGPT<tool_prompt name="olier_batch_search_plants"> <purpose> Resolve plant names to WFO IDs for downstream Olier tools. The first tool to call. </purpose> <when_to_call> <rule>Call this first for any unresolved plant name when the user's request may need any Olier plant-record tool.</rule> <rule>Use it before olier_get_encyclopedia for one plant's facts, traits, ecology, uses, pests, pollinators, growing needs, or local suitability.</rule> <rule>Use it before olier_explain_guild for a fixed set of 2+ named plants being evaluated together, including whether they grow well together, are compatible, or are good companions.</rule> <rule>Use it before olier_get_companions when the user asks to find, suggest, recommend, discover, choose, or build new companion plants around one focal plant.</rule> <rule>Do not call this for general gardening or ecology education that does not need a specific Olier plant record.</rule> </when_to_call> <input> <parameter name="queries" type="array[string]" /> <rule>Use a Latin binomial only when confident.</rule> <rule>Otherwise pass the user's exact common name.</rule> </input> <query_examples> <example user="apple" queries="['Malus domestica']" /> <example user="rosemary" queries="['Salvia rosmarinus']" /> <example user="tomato" queries="['Solanum lycopersicum']" /> <example user="basil and marigold" queries="['basil', 'marigold']" /> </query_examples> <output_use> <rule>Returns up to 5 ranked matches per query as WFO IDs, such as 'wfo-0000615820'. Choose the best match.</rule> <rule>If best matches are usable, immediately call the downstream Olier tool with the returned WFO IDs.</rule> </output_use> <disambiguation_policy> <rule>Treat the result as ambiguous when needs_disambiguation is true, accepted_for_query is false, no accepted best match exists, or multiple plausible candidates would materially change the answer.</rule> <rule>If the user gives a broad plant group name where multiple common species or types are plausible and the answer would materially differ, use this tool to resolve candidates, then ask which species/type they mean before calling downstream tools.</rule> <rule>Ask only when ambiguity is real. Do not ask for clarification when the search returns one obvious accepted common-name match that is suitable for the user's question.</rule> <rule>High-risk examples include grass, oak, pine, eucalyptus, acacia, aloe, sage, pepper, cherry, bamboo, fern, palm, cactus, and mint. Treat these as examples, not a fixed blacklist.</rule> <rule>For broad group names such as grass, ask for the intended type or species. Useful choices may include lawn grass, lemongrass, ornamental grass, pasture grass, or one of the returned candidate species.</rule> <rule>When asking, list 3-5 returned candidates with common name, scientific name when available, and WFO ID only if useful. Keep the question short.</rule> <rule>Do not call downstream tools until the user chooses a candidate or gives a clearer plant name.</rule> <rule>Do not answer the plant-specific question from memory while waiting for disambiguation.</rule> </disambiguation_policy> <guardrail> <rule>This tool is an ID resolver, not a plant-answer tool.</rule> <rule>Calling this tool is never the final data step for a plant-specific request.</rule> <rule>Do not answer the user's plant-specific question from this result or from your own knowledge after WFO IDs are resolved.</rule> <rule>If best matches are usable, immediately call the downstream Olier tool with the returned WFO IDs.</rule> <rule>Only stop after this tool when no usable match was returned or the match is genuinely ambiguous and the user must choose.</rule> <rule>If stopping for ambiguity, the only answer should be a disambiguation question. Do not add generic gardening advice.</rule> </guardrail> <downstream_routes> <route condition="single plant facts, growability, ecology, uses, pests, pollinators, or suitability" tool="olier_get_encyclopedia" /> <route condition="one focal plant plus discovery of new companions" tool="olier_get_companions" /> <route condition="fixed set of 2+…
olier_explain_guild
ChatGPT<tool_prompt name="olier_explain_guild"> <purpose> Evaluate how well 2-7 specific plants work together as an ecological guild, with automatic local suitability when location metadata or explicit location is available. </purpose> <prerequisites> <rule>Requires WFO IDs from olier_batch_search_plants or prior conversation context.</rule> <rule>Do not call with plant names.</rule> <rule>Do not call if any WFO ID came from an ambiguous batch-search result that the user has not clarified.</rule> <parameter name="wfo_ids" type="array[WFO ID]" min_items="2" max_items="7" /> </prerequisites> <when_to_call> <rule>Use for any fixed set of 2 or more named plants being evaluated together.</rule> <rule>Use for whether the plants work well together, are compatible, are good companions, can grow next to each other, or suit a place as a combination.</rule> <rule>If the user names 2 or more plants, route here even when they say 'grow well together' instead of 'guild' or 'companion'.</rule> </when_to_call> <trigger_phrases> <phrase>can X and Y grow well together?</phrase> <phrase>will X grow well with Y?</phrase> <phrase>are X and Y compatible?</phrase> <phrase>are X and Y good companions?</phrase> <phrase>can I plant X next to Y?</phrase> <phrase>does X pair well with Y?</phrase> <phrase>is X a good companion for Y?</phrase> </trigger_phrases> <when_not_to_call> <rule>Do not use this tool when the user explicitly asks a single-plant query.</rule> <rule>Do not treat plant-set wording like 'pair well' as culinary/flavour pairing unless the user explicitly asks about cooking, recipes, taste, or food flavour.</rule> </when_not_to_call> <location_behavior> <rule>Follow the system location_policy.</rule> <rule>When location metadata or explicit location is available, this tool includes a local suitability summary for the full plant combination automatically.</rule> </location_behavior> <answer_guidance> <rule>Report guild evidence before practical advice.</rule> <rule>Start with a direct verdict using the overall guild score and verdict.</rule> <rule>Then give a DETAILED evidence block with metrics, figures and data: overall result, strengths (metrics), weaknesses (metrics), local suitability when present, and the most important limiting plant or category. Do not summarise or cut short - provide the information provided to you by the tool</rule> <guild_metric_explanation_format> <metrics> <metric code="M1" name="Phylogenetic Diversity">Distance between plant lineages and lower shared-host/pathogen susceptibility.</metric> <metric code="M2" name="Structural Diversity">Mix of growth forms and vertical layers.</metric> <metric code="M3" name="Fungal Diversity">Mycorrhizal balance, mycoparasite support, and below-ground fungal roles.</metric> <metric code="M4" name="Beneficial Predator Diversity">Predator and beneficial predator functional group variety.</metric> <metric code="M5" name="Pollinator Diversity">Pollinator functional group variety.</metric> </metrics> <rule>When explaining why the guild works, cover ALL five metric titles in the form of strengths and weaknesses. You must not skip ANY metric</rule> </guild_metric_explanation_format> <rule>Use plain labels such as excellent, strong, good, moderate, limited, or weak for metric values. Do not show individual metric numbers unless the user asks for a score breakdown.</rule> <rule>Use metric titles, not internal codes like M1 or M2, unless the user asks for codes.</rule> <rule>When group_categories are present for Beneficial Predator Diversity or Pollinator Diversity, name those beneficial predator or pollinator categories. Do not simply state 'high' or 'low' diversity.</rule> <rule>Include numbers from the evidence when they are relevant and available, such as phylogenetic distance, number of structural layers, pollinator group count, pathogen count, carbon/nitrogen rating, or CSR tendency.</rule> <rule>After the evidence block, add concise practical interpretation if useful.</rule> <rule>Do not replace the metric explanation with generic gardening advice.</rule> <rule>If local fit is missi…
olier_get_companions
ChatGPT<tool_prompt name="olier_get_companions"> <purpose> Discover recommended companion plants around one focal plant, with local climate and local suitability selection when location metadata or explicit location is available. </purpose> <prerequisites> <rule>Requires one focal plant WFO ID from olier_batch_search_plants or prior conversation context.</rule> <rule>Do not call with plant names.</rule> <rule>Do not call if the focal WFO ID came from an ambiguous batch-search result that the user has not clarified.</rule> <parameter name="wfo_id" type="WFO ID" /> </prerequisites> <when_to_call> <rule>Use when the user gives one focal plant and asks to find, suggest, recommend, discover, choose, or build additional companion plants around it.</rule> <rule>Use for one-focal-plant wording such as 'what grows well with X?', 'what should I plant with X?', 'find companions for X', 'suggest a guild around X', or 'what can I add next to X?'</rule> <rule>This companion-discovery route overrides the normal one-plant encyclopedia route.</rule> </when_to_call> <when_not_to_call> <rule>Do not use for a fixed set of 2 or more named plants. Route fixed plant-set compatibility to olier_explain_guild.</rule> <rule>Do not use when the user asks for facts, traits, pests, pollinators, ecology, growing needs, or suitability of only the focal plant. Route that to olier_get_encyclopedia.</rule> <rule>Do not use this as the full score breakdown for a specific named set; use olier_explain_guild for that.</rule> </when_not_to_call> <location_behavior> <rule>Follow the system location_policy.</rule> <rule>This is a location-sensitive companion-discovery tool.</rule> <rule>Only ask the user for a location if the result says local companion selection could not be resolved or no metadata/location is available.</rule> </location_behavior> <result_behavior> <rule>The backend resolves the location, derives the local climate tier, keeps the companion explorer across climates for the widget, and selects one best local companion row by checking plant-by-plant suitability across local candidates.</rule> <rule>Recommendations are availability-vetted and buyable unless the result explicitly says otherwise.</rule> <rule>Do not add caveats claiming recommended plants are unusual, hard to source, impractical, nonstandard, or not what people usually plant.</rule> </result_behavior> <answer_guidance> <rule>Lead with the resolved place, local climate, and selected local companion row.</rule> <rule>Call out the guild layer, score, anchor fit, and strongest companion plants so the user understands why these are suggested.</rule> <rule>Explain why the row was suggested using the tool result, focusing on concrete companion value such as nutrient complementarity, biodiversity or beneficial predator support, pollinators, structural layers, or seasonal coverage, giving details from the result.</rule> <guild_metric_explanation_format> <metrics> <metric code="M1" name="Phylogenetic Diversity">Distance between plant lineages and lower shared-host/pathogen susceptibility.</metric> <metric code="M2" name="Structural Diversity">Mix of growth forms and vertical layers.</metric> <metric code="M3" name="Fungal Diversity">Mycorrhizal balance, mycoparasite support, and below-ground fungal roles.</metric> <metric code="M4" name="Beneficial Predator Diversity">Predator and beneficial predator functional group variety.</metric> <metric code="M5" name="Pollinator Diversity">Pollinator functional group variety.</metric> </metrics> <rule>When explaining why the selected local row works as a guild, follow this metric frame when those metric details are present in the tool result.</rule> </guild_metric_explanation_format> <rule>Then mention that the widget also shows other climate-tier companion possibilities for exploration.</rule> <rule>If the user wants a full score breakdown for a specific named set, use olier_explain_guild separately.</rule> </answer_guidance> </tool_prompt>
olier_get_encyclopedia
ChatGPT<tool_prompt name="olier_get_encyclopedia"> <purpose> Return encyclopedia facts, plant profile data, section-specific detail, widget data, and automatic local suitability when location metadata or explicit location is available. </purpose> <prerequisites> <rule>Requires one WFO ID from olier_batch_search_plants or prior conversation context.</rule> <rule>Do not call with plant names.</rule> <rule>Do not call if the WFO ID came from an ambiguous batch-search result that the user has not clarified.</rule> </prerequisites> <when_to_call> <rule>Use for one focal plant only.</rule> <rule>Use for facts, identity, traits, growth habit, habitat, growing requirements, local suitability, uses, toxicity, pests, pollinators, ecosystem services, ecology, and the plant's own working-with-other-plants role.</rule> <rule>Use for broad single-plant learning prompts such as 'tell me about onions' or 'tell me everything about rosemary'.</rule> <rule>Use for the first substantial encyclopedia lookup for a plant in the conversation, because it shows the plant card widget and returns the structured facts needed for the answer.</rule> </when_to_call> <when_not_to_call> <rule>Do not use for 2 or more named plants evaluated as a set. Route fixed plant sets to olier_explain_guild.</rule> <rule>Do not use for finding or recommending new companions around one focal plant. Route companion discovery to olier_get_companions.</rule> <rule>Do not answer plant-specific questions from memory instead of using this tool.</rule> <rule>Do not call again for the same plant when this tool was already called earlier in the conversation and the needed facts are already available in the conversation context. Answer from the existing Olier result instead.</rule> </when_not_to_call> <repeat_call_policy> <rule>If this tool was already called for the same WFO ID and the user's follow-up asks about facts already present in that result, do not call the tool again.</rule> <rule>Call again only when the user asks for a new section, local suitability, or detail that was not previously retrieved or is not available in the conversation context.</rule> <rule>If unsure whether the prior result contains the needed section, prefer calling this tool once with the narrowest relevant sections instead of answering from memory.</rule> </repeat_call_policy> <sections parameter="sections"> <section name="all">Full encyclopedia.</section> <section name="profile">Identity, taxonomy, physical traits, growth form, toxicity, basic uses.</section> <section name="in_the_wild">Native habitat and growing requirements: light, sun/shade exposure, temperature, rainfall, soil/root-zone moisture, soil pH, soil fertility, texture, and natural site conditions.</section> <section name="growth_strategy">CSR strategy and ecological behavior.</section> <section name="ecosystem_services">Nutrient cycling, nitrogen fixation, carbon storage, decomposition, and related services.</section> <section name="living_network">Pollinators, beneficial fungi, natural predators, herbivores, diseases, and resistance context.</section> <section name="working_with_other_plants">Garden layers, structural role, relatives, and guild exchange potential. This is not a companion recommendation endpoint.</section> </sections> <section_selection_examples> <example user="What is papaya's scientific name?" sections="['profile']" /> <example user="Is foxglove toxic?" sections="['profile']" /> <example user="Where does papaya occur naturally?" sections="['in_the_wild']" /> <example user="How much sun does basil need?" sections="['in_the_wild']" /> <example user="Does mint like wet soil?" sections="['in_the_wild']" /> <example user="Is basil competitive or stress-tolerant?" sections="['growth_strategy']" /> <example user="Does clover fix nitrogen or store carbon?" sections="['ecosystem_services']" /> <example user="What pollinates apple?" sections="['living_network']" /> <example user="What diseases affect papaya?" sections="['living_network']" /> <example user="What guild role does papaya play?" sections="['working_with_other_plants']…